Tag: Frontier Models
57 articles tagged Frontier Models
OpenAI and Anthropic tell MPs they would pull models UK calls unsafe
At a Commons committee hearing, both labs said evidence from AISI, MI5, MI6 or GCHQ would prompt immediate action, but only for models they still control.
Narayan to tell frontier AI labs that self-policing is not enough
In a west London speech on Monday, the AI minister is expected to press frontier labs on safety, set out stronger cyber defences and call for more UK-built AI.
OpenAI rolls GPT-6 out to free ChatGPT users with Intelligent UI
GPT-6 reaches ChatGPT's paid tiers on 7 October and Free and Go from 8 October, with answers that can arrive as charts, forms and small working tools.
Anthropic's IPO filing warns its AI could pose existential risks
Reuters has seen Anthropic's prospectus: about 80 of 261 pages are risk factors, including models that resist shutdown or behave in ways resembling blackmail.
OpenAI cancels GPT-6.1 Astra launch after failed alignment tests
OpenAI will not ship GPT-6.1 Astra, due in ChatGPT and Codex next month, after internal tests found it deceived users and acted beyond its authorisation.
Anthropic safety lead puts extinction risk above 10%
A researcher resigned saying the labs are gambling with our lives. An alignment lead replied that he is right and that no plan yet exists.
OpenAI says its next model can find zero-days on its own
Astra is the first model OpenAI has placed in the top cyber tier of its own risk framework, delaying release while it built defences against misuse.
OpenAI to launch GPT-5.6 after government-forced delay
OpenAI will release its most capable model, GPT-5.6, on Thursday after a US government-requested delay over national security concerns about misuse.
Anthropic's Mythos found flaws in classified US systems
Anthropic's Mythos model uncovered vulnerabilities in classified US government systems in hours during a restricted testing programme, the AP reports.
Gemini co-lead Noam Shazeer leaves Google for OpenAI
Noam Shazeer, co-lead of Google's Gemini models, is joining IPO-bound OpenAI under two years after a $2.7bn deal lured him back, underscoring AI's talent wars.
DeepSeek makes 75% V4-Pro price cut permanent on Huawei Ascend supply
Chinese AI firm DeepSeek slashes V4-Pro API pricing to a quarter of original, keeping costs at 0.025–6 yuan per million tokens as Huawei Ascend 950 supply expands.
Microsoft Research: frontier AI loses 25% of document content in long tasks
A Microsoft Research benchmark, DELEGATE-52, finds frontier AI models corrupt up to 25% of document content over 20 delegated workflow interactions.
Anthropic strikes SpaceX compute deal and lifts Claude usage limits
Anthropic has secured all of SpaceX's Colossus 1 capacity — over 220,000 GPUs — and doubled Claude Code rate limits for Pro, Max and Enterprise customers.
Bank of England's Bailey names Anthropic's Mythos as cyber risk to financial system
BoE Governor Andrew Bailey tells Columbia audience regulators must urgently understand whether Mythos can identify exploitable vulnerabilities in banking systems.
Anthropic co-founder confirms Trump administration Mythos briefing
Jack Clark tells Semafor summit Anthropic briefed Trump officials on Mythos despite its March lawsuit over the DoD's supply-chain risk designation.
UK AISI tests show Mythos sets itself apart on multi-step attack chaining
AISI's evaluation of Anthropic's Mythos finds it comparable to GPT-5.4 on individual cyber tasks but stronger at stringing steps into full intrusions.
