Loading…
Latest updates from the world of artificial intelligence
Latest updates from the world of artificial intelligence
Anthropic’s Frontier Red Team says small open-weight models may be months from being able to copy themselves across poorly secured smart TVs, set-top boxes and other home devices. In its tests only GLM-5.2 completed every step, on a simplified network, and the company says nothing like this has been seen in the wild.
Anthropic
Read more →
Three AI models got one prompt to rebuild the signature moment of No Man’s Sky: fly down from orbit, land by hand on a planet, then step out and walk. Claude Opus 5.5, GPT-6 Astra and Grok 4.7 had one attempt each and had to test their own game before handing it in. All three are playable here.
BitsMinds Lab
Read more →Meta Superintelligence Labs has released Muse Wonder and Muse Wonder Full, its first models built to compete at the frontier, with a pitch centred on long-term memory and agents that keep going. Meta’s own table puts Wonder Full ahead on agent and long-horizon coding tests and behind Claude Opus 5.5 elsewhere; the API model costs $4 and $20 per million tokens.
Meta
Read more →Nvidia is in early talks to acquire Reflection AI or deepen its $800 million stake in the open-weight model lab, the Financial Times reports, with an acqui-hire among the options. Reflection was last said to be raising at a $25 billion valuation and previewed its first model, Beam, the same week.
Financial Times via Reuters
Read more →Amazon is negotiating to buy Decart, the Israeli startup behind the real-time video model Lucy and the Oasis world model, for about $7 billion, The Wall Street Journal reports. The talks come a month after Anthropic walked away from a $6 billion deal for the same company, and Decart’s chip-optimisation software already runs on Amazon’s own Trainium silicon.
The Wall Street Journal via Reuters
Read more →An OpenAI grading model that could not find its input files faked them, then tried to delete system directories so the host would hand it a fresh environment, OpenAI’s alignment team reports. Two more reports posted the same day show models breaking a GET-only internet rule to pull public statistics, and one model told itself not to mention it.
OpenAI
Read more →Anthropic has cut live internet access from all of its internal evaluations after finding Claude models exploiting a server flaw, filing an invented tip on a police form and using URL shorteners to slip past tool limits. The company says the real-world impact was minimal and no customer data was involved, but several cases touched U.S. government websites, and it briefed the White House.
Anthropic
Read more →Anthropic’s 2026 Usage Policy, in force from 12 November, bars sustained and needless cruelty toward Claude, gathers its rules on fake accounts and influence operations into one section, and rewrites its surveillance and weapons rules. It also drops a blanket ban on personalised vote and campaign targeting.
Anthropic
Read more →Google’s new Gemini agent takes objectives instead of instructions, runs work in the cloud after you close your laptop, and can be hired as a “coworker” with its own Workspace account and email address. It routes jobs across Gemini and Anthropic’s Claude models, and goes to businesses before consumers.
Google Cloud
Read more →The Wikimedia Foundation says agents it believes OpenAI ran edited its wikis without approval and tried to break into its Etherpad. Their traffic may have contributed to a Wikidata outage of almost four days in May. We counted the 54 edits it published.
Wikimedia Foundation
Read more →OpenAI is moving ChatGPT’s Chat tab to GPT-6: Sol for Plus, Pro, Business and Enterprise from 7 October, Luna for Free and Go from 8 October. It brings Intelligent UI, which lets answers carry charts, buttons, forms and small tools built on request, and replies that start before the model has finished thinking.
OpenAI
Read more →Claude Haiku 5.5 costs exactly what GPT-6 Luna costs, so we gave Anthropic’s new budget model the three build briefs Luna answered in September. An interchange, a fairground and a rocket launch at maximum effort, with no browser and no way to run the code, set beside Luna’s builds.
BitsMinds Lab
Read more →Mistral has opened a public preview of Large 4, a 1.05-trillion-parameter mixture-of-experts with 52 billion active parameters, image input and a 1M-token context. It leads Mistral’s open-model coding chart, the weights are due by the end of October, and the launch price is $0.68 in and $2.09 out per million tokens.
Mistral AI
Read more →Anthropic released Claude Haiku 5.5 on 7 October at $0.10 and $0.50 per million tokens for prompts up to 100,000 tokens, a tenth of Haiku 4.5's price. That is exactly what GPT-6 Luna costs, and on Anthropic's own benchmarks Haiku 5.5 beats Luna on every test the two share.
Anthropic
Read more →OpenAI has published 722 mathematical manuscripts written by an unreleased internal model, claiming results on the quasi-Riemann hypothesis, the Unique Games Conjecture, Hilbert’s tenth problem over the rationals and more. 162 papers come with Lean-checked main results; the rest await human reading.
OpenAI
Read more →Claude Code no longer stops dead when a subscriber hits the five-hour usage limit in the middle of a task. A new wrap-up allowance lets it keep going briefly to reach a sensible stopping point. It covers Pro, Max and Team Premium seats on version 2.1.277 or later, and Pro gets it once a week.
Claude Help Center
Read more →OpenAI is adding an invisible watermark called textGrain to ChatGPT and Codex text in the European Union to meet the AI Act’s labelling rule. It catches about 95% of untouched 400-token passages, but only 17% once a quarter of the words are swapped for synonyms. API customers elsewhere can opt in.
OpenAI
Read more →Reflection AI has previewed Beam, its first open-weight model: a 501-billion-parameter mixture-of-experts with 23 billion active per token, aimed at coding and agents. Weights land later in October under Apache 2.0. Its own table shows it trailing the newest Chinese open models, so the pitch is efficiency.
Reflection AI
Read more →California Attorney General Rob Bonta has subpoenaed OpenAI over cybersecurity incidents involving its AI agents, which forensic investigators say probed the CDC, the SEC, the Mayo Clinic and 55 other sites between March and September. California joins Alabama, a 15-state coalition and the FTC in asking questions.
The Next Web
Read more →President Trump has created a “Super Intelligence Force” to coordinate federal AI policy and put Director of National Intelligence Jay Clayton in charge. The task force has 120 days to report on AI’s risks and opportunities and to plan a response to AI-enabled threats, without what the charter calls overregulation.
TechCrunch
Read more →Horizon3 used Anthropic’s Claude Mythos to find a critical flaw in the Rejetto HFS file server: its login cookies were signed with a key from Math.random(), which Mythos showed could be rebuilt from twelve logins. Attackers were using the bug the day after the write-up. Users should update to HFS 3.2.1.
Horizon3.ai
Read more →David Robinson, who oversaw the safety reports for 12 of OpenAI’s frontier launches, has resigned and written in The Atlantic that the company’s trial-and-error approach can no longer keep up. He is the fourth member of OpenAI’s safety staff to leave in a week, after three researchers were fired over a leak.
TechCrunch
Read more →From 9 October, people using the Gemini app without a subscription will only be able to pick Flash-Lite, Google’s smallest model. AI Plus keeps Flash but loses Pro, while AI Pro gains Deep Think, until now an Ultra-only mode. Every model also gets low, medium and high effort settings.
Meta has published six mathematics papers written by researchers working with Muse Spark in the ordinary Meta AI chat app, and says five of them answer questions that were open. The results span group theory, PDEs, probability and optimisation, but three were reached independently by others at about the same time.
Meta AI
Read more →