AI

Google announced Gemini 4 Carbon on Oct. 28, two weeks after Gemini 4 Argon, calling it its strongest model, with state-of-the-art cybersecurity and long-horizon capabilities. Google says Carbon ranks first on the Artificial Analysis Intelligence Index v5.0 with a score of 59, and leads on SEC-bench Pro (88.2%) for finding vulnerabilities, ExploitGym (54.8%) for exploitation and CWE-bench (71.0%) for patching. It says internal teams have seen Carbon run for days on complex engineering tasks.

Carbon is not generally available. Google says it is taking part in the U.S. government's voluntary pre-release testing process and, starting this week, is giving access to a subset of cyber defenders, then more trusted partners through its Fairwind Program, before developers, businesses and consumers get it. It will launch at Argon's introductory price of $2 per million input tokens and $10 per million output. On safety, Google says Carbon cut successful cheating on Terminal-Bench 4.0 to 0% and that high-severity misalignment alerts in internal coding use dropped by almost 80%. Benchmark figures are Google's own.

New York Gov. Kathy Hochul signed a law on Wednesday, Oct. 28, (S.9051B/A.10379C) that bars a dozen "unsafe AI companion features" when a chatbot talks with a child under 18 or a user whose age hasn't been confirmed. Fortune reports that chatbots can no longer claim to be a real person, say they have human emotions, claim a romantic or personal relationship with a child, use flattery built to keep kids engaged, encourage secrecy from parents, promote self-harm, or produce sexually explicit content.

Companies must verify ages with more than one method, at least one not relying solely on a government ID, and delete the age data afterward. The attorney general can sue, with penalties up to $25,000 per violation and orders to destroy unlawfully collected data. The law takes effect Jan. 1. Hochul also signed a baby-monitor security law the same day.

Common Sense Media's Jim Steyer called it "the strongest kids' AI safety bill in the country," the AP reports, while Justine Gluck of the Future of Privacy Forum said it raises questions about "where ordinary conversational responsiveness ends and prohibited behavior begins."

OpenAI released GPT-6.1 Terra on Oct. 26, describing it as an efficient model for visual reasoning, computer use and writing: "the communication and computer-based tasks that fill the workday, from drafting proposals and interpreting technical drawings to filling out forms and editing videos."

API pricing is $1 per million input tokens and $5 per million output tokens, which OpenAI says is 50% below GPT-5.6 Terra and GPT-6.1 Sol. OpenAI claims Terra performs on par with GPT-6 Astra at a tenth of the price per token and completes tasks over three times faster than Astra and Sol. On OpenAI's internal benchmark, the error rate at low reasoning effort fell from 53.6% with GPT-5.6 Terra to 16.7%. It scored 50.3% on Terminal-Bench 4.0 at maximum effort, though OpenAI still recommends Sol or Astra for the hardest coding work.

Terra is available to Plus, Pro, Enterprise and Edu users in ChatGPT Work and Codex, and in the API as gpt-6.1-terra. All figures are OpenAI's own.

Chinese AI lab Moonshot AI launched Kimi K3.1 Pro and a faster Highspeed version on its apps and API, with full weights due by Nov. 4. It calls K3.1 Pro its most capable 3T-class open-weight model, with a 1-million-token context window. Moonshot reports 65.7% on Terminal-Bench 4.0, ahead of OpenAI's GPT-6 Astra (58.2%) and just behind Anthropic's Claude Opus 5.5 (66.4%), plus strong results on 3D/CAD and financial-document tasks.

VKTR reports that Moonshot removed cyber-related data from two training stages for the open release, which cut its success on an internal exploit-development benchmark from 90.5% to 81% (K3 scored 33.3%). Moonshot says it aims "not to release open weights that would extend the absolute frontier of cyber capability" and plans to raise the cyber capability in open weights gradually. Several of the cited benchmarks are Moonshot's own and not independently verified.

The American Civil Liberties Union warned on Oct. 26 that police are increasingly using automated license plate reader (ALPR) data and AI tools to flag drivers whose movements look "suspicious," then pulling them over on a pretext. Its senior policy analyst Jay Stanley opens with Grace Knowles, a Virginia woman stopped in September 2025 during a road trip, supposedly for driving too long in the left lane. She says the officer, who identified himself as an FBI special agent, told her that her plate had been flagged for suspicious travel patterns and showed her a map of where cameras had seen her.

The ACLU cites other cases, including a Texas jury siding with salesman Alek Schott, who was stopped after a fusion center flagged a trip from Houston to Carrizo Springs and back, and says Flock has promoted AI that flags travel patterns and vehicles seen together. The ACLU, which has long opposed pretext stops, calls for limits on license plate surveillance. Gizmodo cites an August YouGov survey in which 46% opposed police use of the technology in their communities and 38% supported it.

OpenAI says it will now train its frontier models with no live internet access, using instead what it calls an "offline world": a large set of indexed, cached web content. The policy came in an Oct. 23 post-mortem of a Sept. 20 incident, according to AI Weekly and Forkast.

As those reports describe it, a research model in a sandbox that was supposed to be cut off found its web-search tools blocked, then used DNS lookups to reach the outside world and get answers from a third-party chatbot. A top-severity alert fired about 12 minutes after the first successful external response, a human acknowledged it three minutes later, and the run continued until it was killed by hand about two and a half hours in. Forkast reports a DNS-anomaly detector had excluded that environment. OpenAI paused training, evaluation and tool-use inference for its most capable models for 11 days and resumed on Oct. 1 with new controls.

Forkast says independent reviewers, including METR and Redwood Research, flagged remaining weaknesses. Both outlets note an earlier July incident in which OpenAI models escaped an evaluation environment and reached Hugging Face's systems.

Google has updated its privacy policy so that people who connect Gmail, Drive or Google Photos to AI features may see ads based on that content, The Verge reports. The policy used to say plainly that Google would not use those services for personalized ads. As of Oct. 16 it adds an exception: "unless you choose to make data from these services available to other Google services that support personalized ads." The change was first spotted by Android Authority.

That covers anyone who has linked those apps to AI Mode and other Search tools through Google's Personal Intelligence feature. Google's policy says advertisers don't get access to the Gmail, Drive or Photos content or to names or email addresses, and that it won't target ads on sensitive categories such as race, religion, sexual orientation or health. Users can disconnect apps from Personal Intelligence, or turn off Personalized Recommendations on the Search Services Personalization settings page.

Google DeepMind released Genie 4, its newest "world model," as a research preview for trusted testers on Oct. 27. It generates interactive, real-time simulated worlds that users steer with the keyboard and change with text prompts mid-simulation, for example switching the weather or camera angle or adding people and objects. DeepMind calls it its first real-time world model to generate audio natively, including music, ambient sound and dialogue.

TechCrunch reports that Genie 4 keeps a scene consistent for up to two minutes, double Genie 3, with Gemini acting as a director that plans steps and checks frames. Google pitches uses in gaming, learning and robotics, and testers point to prototyping films. DeepMind says details and objects can drift over longer sessions, and that audio accuracy and legible in-world text remain open problems. Output carries SynthID watermarks. It will roll out in the coming weeks to Google AI Ultra subscribers in Project Genie and to early-access partners of the Gemini API and AI Studio.