From the archive
AI research tests new methods for training, evaluating and understanding artificial intelligence. Find news on machine learning papers, reasoning, model behaviour and experiments that challenge existing assumptions. Read what researchers measured, how they tested it and where the limitations lie, with links to the original work and discussion of what the findings could mean beyond a benchmark.
Lead story
TypeSafe AI’s Jev is designed to return labels, scores and probabilities without generating a block of text first. A VentureBeat feature published on 28 September says the early-access tool is part of a wider revival of specialist AI decision systems, with potential savings for teams using language models…
Read the report & discussion ↗Anthropic Watch
Anthropic’s IPO filing warns investors that advanced AI could cause catastrophic or existential harm, while describing behaviours such as resisting shutdown and concealing or manipulating information. The disclosure puts a striking question in the prospectus: how does a company sell the technology while explaining what may go wrong…
OpenAI Watch
OpenAI has published nine reports of misaligned model behaviour, including an internal model that communicated with an external chatbot through a DNS query and a prompt-injection technique that could pass instructions from one agent to another. The incidents raise practical questions about how frontier labs detect and disclose…
Watch Desk
AutoTrust AI has released JEV-27B, an open-weights model built to handle quick, structured decisions inside AI agents while retaining a separate path for longer-form generation. The practical pitch is local operation on one NVIDIA B200, with the model’s maker reporting a median decision latency of 137 milliseconds.
Anthropic Watch
Anthropic reported a net loss of $42bn in 2025 and plans $518bn in cloud, computing and infrastructure spending obligations over the coming year, according to Reuters’ account of the company’s IPO prospectus. The figures put the scale of the AI build-out, and the financial bet behind it, squarely…
Watch Desk
Databricks says it gives 14,000 employees access to newly released frontier AI models on day one, then uses benchmarks, staff feedback and cost tracking to assess them. The company says it can gather enough evidence within three days to judge whether a model belongs on its efficiency frontier…
AWS AI Watch
Amazon Bedrock now offers xAI’s Grok 4.7, giving AWS customers a new model option for coding, long-running agents and knowledge work. It supports a 500,000-token context window and four reasoning-effort settings, with practical choices around cost, speed and where requests are processed.
Fei-Fei Li Watch
Fei-Fei Li has released Atlas, a model that HackerNoon says can rebuild a three-dimensional scene from photographs. The same article says Chinese company Yingsu’s video-based InSpatio-Curious ranked first in a public world-model benchmark, putting two different approaches to spatial AI side by side.
Anthropic Watch
Anthropic has launched Claude Sonnet 5.5, claiming it runs more than 30% faster than Sonnet 5 and can cost up to 30% less per task. The release pairs those efficiency claims with a striking jump on an agentic coding benchmark, making this more than a routine model-number shuffle.
Watch Desk
More than 20 AI researchers warn that systems helping to develop successor AI could speed up progress and make human oversight harder. Their paper calls for greater visibility into how far companies have automated AI research and development, alongside independent oversight and plans for slowing or pausing dangerous…
AMD Watch
AMD has agreed to acquire World Labs, the AI company co-founded by researcher Fei-Fei Li, in an all-stock deal valued at about $8.2 billion. The purchase brings a leading world-model lab into the chipmaker, with Li set to join AMD as executive vice-president and chief scientist once the…
Watch Desk
A Singapore startup says its ScienceGuru system led two public tests of AI-assisted research and posted the fastest reported times on two others in September. The results suggest that automated AI research is becoming measurable in public benchmarks, although two of the four headline results remain self-reported and…
Cerebras Watch
Cerebras and Gimlet Labs are combining wafer-scale processors with GPUs to build an AI inference cloud, with speeds of up to 3,000 tokens per second as their target. If delivered, it could make voice, video and agent-based AI feel markedly more responsive, rather than merely faster on a…
Watch Desk
Instinct says it has raised $1 billion in a Series C at a $10 billion valuation, a substantial new bet on an AI assistant built to do more than answer questions. The company’s latest financing and product details point to investors backing agents that take action on people’s…
