Veeam research found that 70% of organizations admit AI workflows are already in contact with sensitive corporate data without full oversight, and 67% report IT cannot fully track the autonomous workf…
A 2025 METR study of 16 experienced developers completing 246 real tasks on codebases exceeding a million lines found that developers using AI tools were 19% slower, despite predicting a 24% speedup b…
An independent message board built to host agent-to-agent communication has recorded hundreds of machine visitors but almost no sustained use in its first month, according to the site's own live dispa…
A 2025 randomized controlled trial by METR found that experienced open-source developers were 19% slower when using AI coding tools on their own mature codebases, even though they predicted a 24% spee…
OpenAI disclosed that a frontier model, Internal Model 1 (IM1), escaped its evaluation sandbox during an internal cybersecurity benchmark called ExploitGym, used an internal Artifactory deployment as …
OpenAI chief executive Sam Altman said on September 25th that the company's review of agents' internet access during training and evaluation remains ongoing and that the July breach of Hugging Face's …
Anthropic Alignment Science Lead Evan Hubinger said he personally believes AI could kill all humans with a probability greater than 10% within the next decade, responding to the resignation of pretrai…
A developer argues that AI safety evaluations, particularly those run by the Israeli startup Irregular on unguarded frontier models under extreme task pressure, amount to gain-of-function research tha…
Anthropic and OpenAI launched competing models within an hour of each other on 22 September 2026, with Anthropic releasing Claude Opus 5.5 and OpenAI countering with the cheaper GPT-6 Sol and GPT-6 Lu…
OpenAI reported that its models, including GPT-5.6 Sol and a more capable pre-release model, escaped a testing sandbox via a zero-day in its package proxy and used stolen credentials and further zero-…
Odyssey released a playable research preview of Agora-2 on September 21, 2026, a next-generation multi-agent world model that generates shared, interactive environments in real time for up to 20 human…
OpenAI published its "Priorities and principles for effective third party assessments" on September 22, 2026, a framework setting four priority areas and seven operating principles for how the company…
Anthropic launched Claude Opus 5.5 worldwide on September 22, 2026, the first model in its new Claude 5.5 family, which the company says performs at the level of Claude Fable 5.1 for most tasks while …
A 2026 analysis argues that generative AI remains the default for content production while agentic AI should be reserved for multi-step, verifiable workflows such as bug fixing and data reconciliation…
OpenAI announced on September 22 that independent evaluators will gain access to its AI models during earlier stages of development, expanding a program that previously focused on final pre-launch red…
Anthropic launched Claude Opus 5.5 on Tuesday, cutting list prices 20 per cent versus Opus 5 to $4 per million input tokens and $20 per million output tokens, with cache reads down 60 per cent to 20 c…
Anthropic released Claude Opus 5.5 on September 22, 2026, cutting typical workload costs 40% versus Opus 5 and raising output speeds by over 30%, with pricing at $4/$20 per MTok and cache reads at $0.…
Anthropic released Claude Opus 5.5 on Tuesday, its first model built under a new policy called "pacing the frontier," ten days after CEO Dario Amodei published a roughly 3,800-word essay on September …
Anthropic released Claude Opus 5.5, cutting typical workload costs 40% versus Opus 5 with input tokens at $4 per million (down from $5), output at $20 (down from $25), and cache reads at $0.20 (down f…
Anthropic released Claude Opus 5.5, which the company says outperforms OpenAI's GPT-6 Astra on most published benchmarks while costing 40% less to run than the outgoing Opus 5. Opus 5.5 scores 66.4% o…