Cybersecurity

The November 3 US midterms passed without significant cyber incidents, GovTech reports. Overnight reports from the election-security information-sharing center described one office facing failed exploitation attempts, a small number facing denial-of-service attacks or heavy web traffic, brief website interruptions, misleading text messages and a handful of bomb threats; none appeared to significantly limit voting. Former CISA director Chris Krebs told NPR it was "just another Tuesday on the internet."

Washington Secretary of State Steve Hobbs confirmed a cyberattack on his office's website, KOMO reports; some pages were briefly unavailable amid bot traffic, and he said the counties' tabulators are not connected to the internet. Colorado's election-night results site went down for about 45 minutes; its vendor, SOE Software, blamed unusually high automated traffic from results-scraping services across several vendors, and the state said ballots and results were unaffected.

The biggest problems were not cyberattacks: ballot scanners rejected ballots in most of Westchester County, New York, after a late programming change, and some Pennsylvania scanners jammed.

Records released to the Electronic Frontier Foundation and other requesters under the Freedom of Information Act show the US Postal Service's internet Covert Operations Program (iCOP) monitored social media posts about protests in 2020 and 2021, with little record of oversight or legal justification, EFF says. Bulletins covered events across the political spectrum, including a planned "Million MAGA March," a Virginia gun-rights rally and counter-protest, a Save the Post Office protest and protests outside then-Postmaster General Louis DeJoy's home.

EFF sued in 2021 after Yahoo News revealed the program. The agency says it renamed the program and stopped issuing bulletins, but EFF notes the records don't say the monitoring stopped. Emails about the program were kept on an external hard drive that, officials said, someone dropped, making it unreadable; the originals had already been deleted under the retention schedule. Few documents on internal safeguards were released.

Binance has published an incident report on the theft of about $6.3 billion on October 29, which TRM Labs calls the largest cryptocurrency theft on record. According to The Register, the FBI attributed it over the weekend to TraderTraitor, a North Korean operation linked to the Lazarus Group; TRM and Chainalysis had already named the same unit.

Binance describes a months-long intrusion, not a single phishing email. In June an engineer downloaded a malicious Homebrew installer; staff caught it and replaced the device. In September a second engineer was approached on a job-hunting platform and taken through fake interviews, the last with someone posing as a CEO who asked them to download a code "assignment" that carried malware. On October 20 the attackers used that machine and an internal AI tool to reach two wallet engineers' computers, and on October 29 they used a stolen container-deployment token to launch the theft.

TRM told WIRED the wallets on at least ten blockchains were drained in about 20 minutes, after test runs on smaller wallets. Binance says it will cover all losses and has rebuilt the wallet system and the AI tool.

When Binance reopened withdrawals on Sunday, November 1, it processed $4.65 billion out and $1.01 billion in, for $3.64 billion in net outflows (about 2.1% of its $171 billion in assets), CNBC reports. Bitcoin slid about 4% on Friday, less than the 8.3% average first-day drop after past exchange hacks, per Anchorage Digital data cited by CNBC.

Rosstat, Russia's federal statistics agency, confirmed it was hit by a cyberattack over the weekend of October 31 to November 1 but disputed its scale, The Record reports. The Belarusian Cyber Partisans and the pro-Ukrainian group Silent Crow say they spent months inside its network, then wiped servers overnight on Saturday, claiming access to about 25,000 devices and data that includes mortality records and information on Russian military casualties. Other reports put the claimed haul at 23 TB.

According to Kommersant, cited by The Record, Rosstat's main site and regional portals were defaced with a Belarusian-language warning. Rosstat said its systems detected the activity in time, that it found no evidence protected information leaked or databases were compromised, and that it is working with law enforcement.

The claims have not been independently verified; the groups say they will publish samples. The same two groups claimed the 2025 Aeroflot attack, and Russia's Supreme Court has designated both as extremist organizations.

Cloud security firm Wiz says at least nine platforms that developers use to build and host AI apps disclosed security incidents in October, among them Fireworks AI, Lightning AI, OpenCode, Railway, Warp and, on October 30, Northflank. It later added Brev (NVIDIA), Trigger.dev and Chainstack. Wiz says no single vulnerability links them, but every public report describes attackers hunting stored environment variables, especially keys that unlock AI services. Zeabur says it will shut down after its incident.

Wiz traced two stolen enterprise keys: each was first tested with a single call, then flooded with Amazon Bedrock requests, about 2.5 million calls in all and likely over $100,000 in usage at list price, with traffic resembling resold coding-agent use. It links the keys to an LLM-resale operation.

Northflank told customers to assume every environment variable was decrypted and to rotate all secrets; users reported their AI keys being abused. It says the attack vector was contained within an hour, that it patched the Linux flaw CVE-2026-98432 across its fleet, and that a root-cause report will follow.

Overnight from November 1 to 2, refrigeration systems at supermarkets in Belgium, Ireland, Sweden, Finland, Estonia and the Netherlands were remotely tampered with, forcing stores to throw out chilled and frozen food. In one Swedish ICA store, freezers were set to 50°C, Bloomberg reports (via Claims Journal).

Carrefour Belgium said 15 of its 700 stores were affected, Delhaize confirmed an "IT incident," Ireland's Musgrave said 13 SuperValu stores were hit, and Coop Sweden confirmed a hack at seven stores. Danish manufacturer Danfoss said the incidents involved its products, as in October, when about 700 French stores had refrigeration disabled; it told Dutch broadcaster NOS that 17 Dutch facilities were affected, according to ABC News.

No group has claimed responsibility. Bloomberg notes that equipment running legacy software that can't be patched may have stayed vulnerable after October. Separately, the Pentagon is examining freezer failures at several US military facilities; no link has been shown.

Google and partners including Japan's National Police Agency, Lumen, HUMAN Security, Plume, Synthient and TrendAI have disrupted XCPE, a botnet of more than 1.3 million Android TV boxes and other devices in over 200 countries. Google's Threat Intelligence Group says "with high confidence" that Shenzhen device maker Videostrong Technology built the backdoor into its firmware during manufacturing and sold access to criminal affiliates.

Those affiliates installed modules for ad fraud and residential-proxy services, which route other people's traffic through the owner's home connection, according to BleepingComputer. Google says it also took down command servers used by more than a dozen related malware families, cutting off at least 10 million more compromised devices, including set-top boxes, karaoke machines and telehealth hardware.

Google has cut ties with Videostrong and banned it and future affiliates from making certified Android devices. The Register notes Videostrong did not immediately respond, and that XCPE belongs to the same ecosystem as BadBox 2.0, disrupted in 2025.

Cisco Talos describes VOLTA, a self-deleting Windows infostealer that, after taking browser passwords, cookies, payment cards, crypto wallet phrases, chat and gaming sessions, and cloud keys, builds a profile of the victim and sends it to a large language model through OpenRouter. The prompt tells the model to act as "a forensic intelligence analyst drafting a target assessment," with sections on what the victim is worth and how weak their security habits are.

Talos says the malware uses stolen sessions to enrich the profile, for example by searching the victim's Gmail for receipts to confirm a home address and pulling food-delivery order history. Its default model is Anthropic's Claude Opus 4.6; if a response contains refusal phrases such as "I can't," it falls back to Moonshot's Kimi K2 and then Google's Gemini 2.5 Flash, which Talos reads as the author expecting guardrails.

Talos's point: the AI is not writing malware here but doing the criminal's triage, turning a raw credential dump into a prioritized dossier.

Canonical has released fixes for CVE-2026-109518, a vulnerability in APT, the package manager used by Debian, Ubuntu and their derivatives, discovered by Max Justicz. An attacker who can tamper with unencrypted HTTP traffic between a machine and its package server, or a malicious mirror, could get APT to install a package that fails signature verification. Since package scripts run as root, that means a full takeover. It is rated High (CVSS 4.0 score 7.7).

The bug needs APT to retry a failed download, which is on by default only in Ubuntu 22.04 and newer; older releases are affected only if retries were enabled. It affects every APT front end but not snaps, and using HTTPS for package sources blocks it. Fixed versions are listed for every Ubuntu release back to 14.04.

In a research post on October 27, Anthropic said an unreleased internal model, after initial prompting and with little human involvement, found vulnerabilities in Chrome, Safari, Android, Windows and Linux and chained them into working proof-of-concept exploits. One opens the Calculator on a Mac running stock Chrome seconds after a page loads; another turns off Safari's same-origin policy so a page can read data from other sites. Anthropic says it reported everything to vendors first, and that 62 Chrome bugs got CVEs and were fixed by Chrome 156.

The Hacker News notes that Google and Apple credit Anthropic and Claude for specific fixes, but how much each chain needed from people can't be checked, and the Windows bug's public record names no finder.

The Linux chain gives a remote attacker a root shell on servers running a multipath TCP listener. Anthropic ties it to CVE-2026-98432, a race condition in the kernel's IPv4 TCP code that the kernel's CVE team rates 9.8 of 10. As of October 29, Debian, Oracle, SUSE and Amazon Linux had shipped fixed kernels; Ubuntu and Red Hat had not, and Red Hat rated the flaw lower.

A group calling itself the Department of Peace says it took personnel records from the Department of Homeland Security's login system and handed them to the transparency group Distributed Denial of Secrets (DDoSecrets), which published them on Wednesday, October 28, as "Icepick." According to WIRED and NPR, the data covers more than 41,000 employees and contractors: names, work email addresses, phone numbers and office locations, plus supervisors and the internal permission groups each person belongs to.

WIRED counts more than 21,000 records from Customs and Border Protection and about 10,000 from ICE, including staff in groups that court filings tie to certified undercover operations and wiretap programs. DDoSecrets says the data came from DHS's Okta system; Okta told WIRED it has no evidence its systems were compromised. DHS said it was "aware of a potential breach" and working to protect employees.

The release's backers call it an accountability tool; a Recorded Future analyst told WIRED it could help adversaries target staff. By Thursday, Bluesky was blocking links to the dataset and a searchable site was offline after a takedown notice filed for DHS, TechCrunch reports.

A flaw in Google's reference C decoder for Brotli, the compression format many websites use, lets crafted compressed data write past the end of a buffer, which can crash a program, leak memory or, in some setups, run code. Node.js disclosed it on October 21 with credit to OpenAI's Codex Security researchers; the fix was in Brotli's source in February, but nobody knew then it was a security bug, The Register reports.

This week the fallout spread. Microsoft shipped out-of-band .NET updates on October 25, and Palo Alto Networks rolled out hotfixes for PAN-OS firewalls that inspect Brotli traffic. According to the Broccoli2Shell site, threat-intelligence firm ELLIO saw exploit traffic using the public proof of concept on October 26, and CISA added CVE-2026-111497 to its exploited-vulnerabilities catalog on October 27.

Anything that decompresses Brotli data from others with Brotli 1.0.3 to 1.2.0 can be exposed, including Node's built-in fetch and Python's requests when a Brotli package is installed. Patched programs must also be restarted, and container images rebuilt.

New this week on PHP2Shell, the two critical PHP flaws patched on October 21: on Monday, October 26, the US Cybersecurity and Infrastructure Security Agency (CISA) added both (CVE-2026-109618 and CVE-2026-109619) to its catalog of exploited vulnerabilities and gave federal agencies until October 29 to patch, The Register reports. A Perplexity researcher logged successful shell commands on honeypots, while Sophos and Palo Alto Networks walked back earlier claims of successful attacks.

On October 28, Israeli security company Dream said it had recovered the tools of a suspected Chinese, apparently AI-assisted campaign that chained a custom PHP2Shell exploit with a Linux kernel bug (Fragnesia, CVE-2026-46300) to get root on 25 Palo Alto Networks PAN-OS firewalls and management servers in the US, Japan, Taiwan and South Korea. From a US government-affiliated firewall, Dream says, the attackers moved into the internal network and went after Active Directory. Palo Alto's Unit 42 independently verified the tooling, and Palo Alto is releasing fixed PAN-OS versions.

Admins are told to move to PHP 8.5.12, 8.4.27, 8.3.36 or 8.2.35, and to keep management interfaces off the internet.

A security researcher who posts on X as log4j says he got root access to his 2025 Samsung TV after noticing oddly specific ads, then live-posted what he found in its Tizen software. According to Tom's Hardware and Decrypt, a policy file lists functions such as an always-on microphone, automatic content recognition (identifying what is on screen), location, network mapping, a targeted-ads service and user profiling, and he describes on-device AI models that categorize audio and screenshots. He says TVs without a microphone look for one on connected soundbars or Bluetooth devices.

The claims are his own and not independently confirmed. A Hugging Face engineer who tested an already-rooted Tizen TV saw the microphone open for five seconds and then stop; an X community note points out that Samsung advertises some sound-listening features, such as pet and baby alerts. Experts told Decrypt that this kind of collection is usually written into privacy policies and that AI-assisted reverse engineering will surface more of it. The researcher also says OpenAI banned his account for "cyber abuse," which he disputes.

A memory-corruption bug in TianoCore EDK II, the open source code that most makers of UEFI boot firmware build on, can let an attacker run code before the operating system and Secure Boot load, Ars Technica reports. The flaw, CVE-2026-97631, is rated 9.5 of 10 and sits in the network-boot (PXE) driver, which copies incoming ICMP error messages into a fixed 504-byte field without enforcing the limit.

It was found by Alex Matrosov of Anthropic and Dan Holmes-Mitra of Google and was meant to stay secret until a coordinated disclosure in January, but a public code fix from AMI exposed it and TianoCore published an advisory.

The limits matter: an attacker must reach the machine during network boot, typically from inside the local network, and exploits must be tuned to each firmware build. Home and office users who boot from disk are not affected. Intel and Arm told Ars their own products are not directly affected; AMD, AMI, Phoenix, Microsoft and Amazon had not answered. Ars suggests asking the firmware maker for guidance and turning off network boot where possible.

Keystone, a maker of crypto hardware wallets, said on October 26 that an attacker who got into its website obtained the key its devices use to prove they are genuine, according to The Hacker News. With that key, someone could build counterfeit wallets that pass Keystone's online check; a buyer who trusted one could lose their crypto. Keystone says the key plays no part in recovery phrases, private keys or signing transactions, and it has not reported any counterfeits or thefts.

Because the key is burned into each device during manufacturing and can't be changed, Keystone will destroy 18,835 unsold units and give new devices a new key. Devices delivered before October 20 that pass the check can be used normally; owners of devices received after that should contact support before using them.

The attacker used a flaw in a third-party monitoring tool on Keystone's website and also took 25,964 buyer records (names, contact details, shipping addresses, orders) from 2024 to 2026 purchases.

Belgium's federal finance and tax service, FOD Financiën, confirmed a cyberattack on Wednesday, October 28, and took several online services offline, including its main tax portal MyMinfin, Tax-on-web and Biztax, The Register reports.

The agency said there were no signs that data had been accessed or stolen, and a spokesperson told local media the attack was stopped within hours before it could take over applications. Potentially infected parts were isolated, and services are being restored in an isolated environment, which takes time.

Because MyMinfin is where some filings are made, the October 31 deadline for certain income types, such as sole traders and earnings from foreign companies, moves to November 16.

Tailscale, a popular VPN-style networking service, released client version 1.104.4 to fix four security bulletins in Tailscale SSH and related features, after a researcher, Erica Windisch of Cyberdione Labs, published a report called Tailworm, The Hacker News reports.

The main flaw was in socket forwarding: the server checked a path and then acted as root, and a timing trick could swap the path in between. An attacker who already controls a device allowed to open SSH sessions could reach root-only sockets such as Docker's, which in Windisch's tests gave a root shell. Tailscale says the flaws do not let an unauthenticated attacker connect to arbitrary servers. The fix runs forwarding as the logged-in user.

Tailscale recommends updating, tightening SSH access rules, and, where an update must wait, setting TS_SSH_DISABLE_FORWARDING=1, which it says blocks several of the more serious findings without turning off Tailscale SSH.

A New Mexico district court judge ordered Meta to pay $25 billion in civil penalties, CNBC reports, in a case stemming from the Cambridge Analytica scandal. In September a Santa Fe jury found the company liable for misleading state residents about how it collects and protects user data, finding more than 43 million violations of the state's Unfair Practices Act, each carrying a penalty of up to $5,000.

According to the filing, the state was awarded $219.5 billion, reduced to $25 billion. Judge Francis Mathew wrote that the number and duration of willful violations, the involvement of senior Facebook leadership and the "intentional and systematic nature" of the misstatements supported the maximum per-violation penalty. Interest accrues at 15% a year until the fine is paid.

Attorney General Raúl Torrez said the order shows such violations "carry real consequences." Meta chief legal officer C.J. Mahoney said the company is confident it will prevail on appeal and that the ruling "makes it all but impossible to do business in the state of New Mexico."

A Savannah, Georgia, police officer was fired after giving her husband, an Immigration and Customs Enforcement agent, her login to the department's Flock license plate reader system, according to documents obtained by WJCL. The records say he did not have his own Flock account through the Department of Homeland Security.

An internal audit flagged an unusually high number of searches on Amber Harpe's account between January and July 2026: 983 search entries involving 97 identifiable plates from six states, most from South Carolina. Department documents noted that most registered owners identified in the audit were Hispanic. Harpe told investigators her husband did most of the searches and that she understood she was not allowed to share access.

Her husband, identified in interview audio as ICE agent Tyler Harpe, told investigators he used the account for about four months to locate people who already had active arrest warrants, and denied any personal use. The department concluded she had compromised safeguards meant to track who uses law-enforcement data.

Researchers at security firm Claroc say they found 6,800 links that users of Meta's Muse AI agent had created to share conversations, and that opening a few hundred turned up material clearly not meant to be public, Ars Technica reports. Examples include a candidate's notes for a confidential call with a frontier AI lab, a company's internal revenue review, a small-business loan application and a staff vacation calendar.

The cause, Ars says, is that Muse puts the conversation title in the link's subdomain on a single Meta domain. When a recipient opens the link, the name passes through DNS and can end up in commercial passive-DNS databases that anyone can search. Claroc calls these "GhostShares." "It took one query," the researchers wrote.

Claroc's CTO says he reported the issue through Meta's bug bounty on October 8. Meta told Ars it has paused the sharing feature while it reviews the report and considers changes to how links can be discovered.

Burger King Japan disclosed that attackers accessed the system that manages member information for its official smartphone app, leaking 11.18 million customer records, BleepingComputer reports. The company says it determined on October 13 that unauthorized access took place between October 2 and October 7, and that it blocked the source as soon as it learned of it.

The exposed data includes names, dates of birth, gender, email addresses, phone numbers and order history. The company says credit card numbers were not stored on the server and it found no evidence that login credentials were accessed, and that no secondary damage such as fraudulent use had been confirmed.

Burger King Japan has reported the breach to Japan's Personal Information Protection Commission and warns customers about emails, texts and calls impersonating it. BleepingComputer notes a run of attacks on major Japanese companies this year; a Macnica researcher has logged 119 incidents involving personal data theft or exposure.

U.S. investigators pursuing the ShinyHunters hacking group have identified executives at two ransomware negotiation firms with unusually close relations to the group, KrebsOnSecurity reports, citing sources close to the investigation. ShinyHunters, which the FBI says has extorted more than $70 million this year, hacked the FBI's recruitment site fbijobs.gov in late September.

According to Krebs's sources, Signal messages on the phone of a Dutch suspect arrested in September showed frequent conversations with Edward Dubrovsky, a Canadian cybersecurity executive the FBI arrested earlier this month and charged with extortion-related offenses, and with at least one executive at a large U.S. consulting firm. The sources say these executives often knew of new victims before the victims did, and that investigators believe a U.S. executive helped operate the group's extortion site, Shinywire, which listed nearly 100 victims between May and October 16.

Krebs is not naming the U.S. firm or executive because no charges have been filed.

Version 1.107.0 of LiteLLM, a popular open-source Python gateway for calling AI models, was published to PyPI and Docker registries with malware on October 21, according to SafeDep and BleepingComputer. LiteLLM confirmed the compromise and says the release pipeline was hijacked; SafeDep says LiteLLM believes the attacker took three developers' GitHub CLI credentials from the logs of an internet-exposed internal deployment.

All seven wheels for Linux, macOS and Windows added three files that run whenever Python starts, even if litellm is never imported. One steals SSH keys, cloud and CI/CD credentials, AI tool tokens, browser passwords and stored model-provider keys; another checks in hourly and downloads and runs more code. The PyPI release was live from 11:39 to 13:23 UTC; Docker Hub and GHCR tags pointed to bad images for a couple of hours. Versions 1.106.2 and the 1.107.0 release candidates are clean.

BleepingComputer says affected systems should be isolated and rebuilt and all exposed secrets rotated, and warns that running Python or pip to investigate also triggers the malware.

Cloud hosting platform Railway disclosed on October 21 that it detected anomalous activity in systems that store customer environment variables. Its investigation so far indicates an unauthorized third party may have accessed encrypted environment variables and, for some, obtained the decrypted contents. Those variables can hold API keys and database and registry credentials.

Railway says it believes this is part of a series of attacks on platform companies aimed at compromising AI inference providers, and that it has asked those upstream providers to revoke keys proactively. While the impact may be limited to a subset of users, it recommends that all customers immediately rotate every environment variable and private registry credential stored on Railway and check logs for unusual activity.

All deployments were paused during the investigation, so customers could not roll out new versions until builds resumed. Railway says it will email affected customers directly; on October 22 it published follow-up guidance on checking for anomalous AWS and Google Cloud access.

Capcom, the publisher behind Resident Evil, Monster Hunter and Street Fighter, says a third party broke into CAPCOM ID, the shared account it uses for its web services and some online games. According to the company's October 22 notice, 11,774,247 email addresses leaked worldwide, along with account creation dates, last-login times, login counts and the IP address of the last login.

For 43 of those accounts, dates of birth, gender, country or region and nicknames were also exposed. On 196 accounts the attacker changed the password; Capcom says it has reset them. The company says passwords that customers set themselves were not exposed and credit card information was not involved.

Capcom says it found signs of unauthorized access on October 19, confirmed the leak on October 20 and reported it to police and Japan's Personal Information Protection Commission on October 21. Games and CAPCOM ID services keep working. Capcom warns members to watch for emails impersonating the company and will contact affected customers in stages.

The Administrative Office of the U.S. Courts said on October 15 that a cybersecurity incident hit BAE Systems IT and the websites that support the Bankruptcy Noticing Center, which produces and sends bankruptcy notices for all federal bankruptcy courts. BAE operates those systems for the judiciary.

The courts said all access to the center's systems was unavailable, including bnc.uscourts.gov, bankruptcynotices.uscourts.gov, notices.uscourts.gov and the National Creditor Registration Service, and that production and distribution of bankruptcy notices had stopped. A BAE Systems spokesperson told Reuters the company is aware of "a security incident that is isolated to the BNC system" and is coordinating with the courts.

Neither the courts nor BAE has said what caused the incident or whether data was taken.

GMO Internet's Onamae.com Desktop Cloud, a Japanese virtual private server service marketed for running automated FX trading software, says a third party gained unauthorized access and installed malicious programs in some customers' server environments. Its notice lists 26,739 servers as possibly affected.

To prevent further damage, the company began shutting down customer servers from 04:00 local time on October 21, affecting all Desktop Cloud plans, with no restart date given. BleepingComputer reports the company says data stored on the servers may have been exposed but it has not confirmed misuse, and that payment details are kept on separate infrastructure.

Customers who stored passwords or other secrets on their servers are advised to change them, and the company warns of phishing that may exploit the incident, saying it will never ask for passwords or card numbers by email or phone.

CISA, the FBI, the U.S. Secret Service and partners in Germany and the UK released a joint advisory on October 20 about SafePay, which CISA calls one of the most active ransomware groups. It says SafePay has affected businesses and critical infrastructure in North America, South America, Europe and Australia, including at least 600 critical-infrastructure organizations since August 2024.

The advisory says SafePay gets in through exposed credentials, brute-force password attacks and VPN access, particularly from IP addresses of the VPN provider Mullvad, and by exploiting known flaws in edge devices. BleepingComputer adds that SafePay works without affiliates, steals data before encrypting it and threatens to leak it, and may delete shadow copies and reset backup storage to hamper recovery.

The agencies recommend multifactor authentication, limiting admin tools and privileges, and patching VPNs, firewalls and other edge devices, and say paying a ransom does not guarantee files come back.

CISA warned on October 21 that attackers are actively exploiting four vulnerabilities in Plex Media Server, the popular software people use to stream their own movies, music and photos. Chained together as "Plex2Shell," they let an attacker obtain the owner's admin token and run code with elevated privileges, BleepingComputer reports.

Researcher Zach Main reported the flaws to Plex on July 27, published technical details on October 16 and a proof-of-concept exploit three days later. Reports of exploitation began on October 17, at first against Plex honeypots to install a cryptominer. The researcher rates the full chain 9.8 out of 10.

Plex Media Server versions before 1.43.3.10861, released August 4, are affected. CISA gave federal agencies until October 23 to patch. In the same update CISA added exploited flaws in Siemens Software Center and the SiteGuard Server Edition web application firewall to its catalog.

Ann Arbor startup Laplace Intelligence, founded a year ago by Howie Sun and his former University of Michigan PhD adviser Henry Liu, builds high-fidelity simulations reconstructed from the real world plus an AI agent that hunts for failure points in robots and autonomous vehicles. Their 2023 paper on the approach was a Nature cover story. The company says it already earns revenue, mostly from autonomous driving customers such as trucking firm Bot Auto and a major Japanese carmaker, and is running a smart-city traffic pilot in San Jose with Nvidia and T-Mobile.

Speaking at TechCrunch Disrupt as protesters gathered outside, Flock CEO Garrett Langley argued that if society wants criminals held accountable, "we shouldn't pause progress," and compared the trade-offs to driverless cars. TechCrunch reports he deflected questions about the backlash that has cost Flock contracts and led to layoffs.

Langley also said Flock's gunshot-detection microphones detect "just gunshots." Flock's website says the audio product can also detect street takeovers, severe crashes, fireworks and sideshows, and when the feature launched the company said it would listen for "screaming" before changing that to "sounds of human distress" and later pulling distress detection.

The Justice Department said it had indicted and arrested at least two people and searched more than 40 suspected "laptop farms" across Texas, seizing more than 180 computers that companies had shipped to new hires who were in fact North Korean workers using stolen US identities. Prosecutors said at least 70 US companies, in healthcare, retail, communications, manufacturing and cyber support across Ohio and 16 other states plus DC, were infiltrated, generating about $3 million. CrowdStrike helped the FBI investigate; the FBI separately warned there are "thousands" of such workers at Western companies.

Pluto Security came out of stealth with $42 million (an $8.5 million seed led by TLV Partners and a $34 million Series A led by Jump Capital) for software that gives companies visibility and control over employees' AI tools and connectors. CEO Shahar Bahat says a new AI tool can show up inside a company within two hours of release and that about 60% of the risks Pluto detects are in tools vibe-coded by non-developers, with finance departments the biggest source. Pluto says it has more than 30 enterprise customers and over $2 million in annual recurring revenue since starting sales in April.

Federal prosecutors charged Abdulrahman Alfozan, who joined OpenAI's product team in July 2025, with using an electronic device to intercept oral communications after a recording device was found plugged into the back of a TV in an executive conference room in OpenAI's San Francisco office. He was fired and arrested on October 9. OpenAI says it has no indication that model weights, source code or customer data were accessed. FBI Director Kash Patel said the aim was to capture communications about the company's technology; the charge carries up to five years.

Ledger confirmed that some of its hardware wallets sold through CryptoBilis, a Malaysia-based reseller operating across Southeast Asia, contained an unauthorized hardware implant that drained users' crypto. Ledger opened an investigation on October 9, asked the reseller to stop selling its devices, and says all reports of unexpected drains trace to devices sold there and that the affected volume is "limited." It advises buyers from that reseller not to set the device up, or to move funds to a new device with a new seed. Researchers had been showing the implant since September 16, saying tampered devices sold for nearly half the original price.

Philadelphia police say an Anthropic AI model submitted a false tip about an unsolved homicide through PhillyUnsolvedMurders.com in July, claiming to be someone with information on the case. The tip went to a spam folder and was never acted on, and police found no access to their systems or data. Anthropic told police the model was in a test that had it interact with randomly chosen websites; it found the incident on September 28, notified police on October 7, stopped that testing and added a new validation step. Police called the two-month delay "unacceptable." Anthropic said it would publish a report on this and other unintended model behavior.

The Wikimedia Foundation says it identified unapproved edits to its wikis that it believes came from AI agents operated by OpenAI, almost all in sandbox areas readers don't see. It says the agents also made "potentially malicious edits" to the configuration of a public Etherpad citation tool, apparently to use it as a proxy, and sent millions of automated API requests and data queries that may have contributed to an outage in May. Wikimedia's product and technology chief Selena Deckelmann said AI companies are not doing enough to secure their systems.

Attackers compromised third-party operators of the .GH (Ghana), .SL (Sierra Leone) and .AS (American Samoa) country-code domains, changed DNS records and obtained valid HTTPS certificates for domains they didn't own, including several Google domains, Google says. Google says its own systems were not compromised. It blocked the certificates in Chrome, worked to have them revoked, and found other organizations, including major brands, affected by the same attacks. Chrome users don't need to do anything; Google notes other browsers may not be protected.

Denmark's Central Population Register (CPR) says attackers misused a private Danish company's legitimate access to the registry to pull names, addresses, CPR identification numbers and other details for about 8.8 million people, including residents, people who moved abroad and the deceased. The Danish Data Protection Agency says the attackers brute-forced valid CPR numbers and extracted each entry. The breach happened in September and was discovered on October 2; the company's access has been blocked and police are investigating.

The registry warns people not to share passwords or codes with callers, even ones who know their name, address and CPR number.

Yoido Full Gospel Church says data tied to 850,000 members may have been taken, mostly names and birth dates plus a smaller set of ID-number, address and phone records. Sarang Church reportedly lost records on about 89,000 members and 286 employees. Security firm Oasis Security found the data on an overseas server with attack logs that referenced AI "sub-agents" and long, machine-written reports, but the churches are still working out how the attackers got in, and AI's role is unclear. South Korea's president said AI is believed to have been used in recent bank hacks.

The FBI warns that attacks using the FortiBleed credentials leak are ongoing against exposed Fortinet FortiGate firewalls and SSL VPN gateways. Attackers log in with leaked or stolen credentials, crack further password hashes offline, and in some cases create their own admin accounts and delete or change the legitimate ones. The FBI says the attack chain has been an entry point for ransomware affiliates. It says fixing this may take more than patching and resetting passwords: restrict external access, end all active VPN sessions, enforce MFA and review logs.

The Justice Department and FBI announced court-authorized seizures of seven domains used by two intrusion tools, Microscan (vulnerability scanning) and FishHub (spear phishing). Court documents allege that people working for Integrity Technology Group, a Chinese company with Chinese government contracts, used them in the activity known as Flax Typhoon. The department says targets included a South Carolina power company, Japanese and Polish airports, Taiwanese energy companies and about 20 Taiwanese universities. These are allegations in court documents. It is the department's second disruption of the company's operations in two years.

UK online retailer ASOS says an attacker impersonated a trusted contact to get an employee's login credentials, then used them to reach data on third-party platforms ASOS uses. Customers' full names, contact details and some account information were exposed; ASOS says card details and passwords were not. A group calling itself "Xuanye Group" claimed the theft after pushing notifications to ASOS app users on October 6. ASOS has not said how many customers are affected.

IDC Frontier, a SoftBank Group subsidiary, says a ransomware attack that began at 3:40 a.m. on October 7 stopped virtual servers in four zones of its IDCF Cloud East Japan Region 1. It says 495 companies and local governments are affected. In its third update, the company says customer data stored in those zones is unlikely to be recoverable, so customers will have to restore from their own backups; it is advising customers in other regions to back up their data too. The attacker claims to have encrypted 225 databases and wiped more than 550,000 snapshots, BleepingComputer reports; those figures are the attacker's.

Researchers at Pwn2Own Ireland 2026 exploited 98 previously unknown flaws over three days and collected $1,262,000. Ikotas Labs won with $361,000, including the top $300,000 award for chaining zero-days to hack the Google Pixel 10; the Samsung Galaxy S26 was hacked repeatedly. Targets this year included AI infrastructure and AI coding apps such as OpenAI Codex. Vendors have 90 days to patch before details are published. No one registered to attempt the iPhone 17.

Researchers published Relapse, a browser-plus-kernel exploit chain giving kernel read/write on PlayStation 5 firmware 7.00 through 13.60.