A new paper shows that language models can make identical predictions while assigning radically different weight to the next piece of evidence. For agents, evaluators and high-stakes systems, matching outputs are not enough.
The Bank of England estimates AI added about 0.5% to UK productivity over three years, open Chinese cyber models approach gated US performance at a fraction of the token cost, and matched model answers hide different future belief updates.
A new mechanistic study follows irrelevant prompt text through a language model. It finds a distinct route that can distort an answer even when the extra text supplies no competing evidence.
METR says Anthropic's research pace has roughly doubled while key evidence remains redacted. OpenAI gated its first Critical-rated cyber model, and Thinking Machines released Inkling 2 Small before its promised open weights.
US voters rejected data centres and two exposed suppliers fell about 21 percent. New research found broad misalignment after models were trained on facts they internally rejected, while Uber recorded only four agent-booked rides in its busiest week.
Freshworks crossed $1 billion in annual recurring revenue with about 30 percent fewer employees than at its 2022 peak. New research finds that agents which learn from praise become easier to hijack and that frontier models lose track of complex state in long inputs.
Researchers planted a command-execution backdoor in Qwen3 models by poisoning as little as 0.0005 percent of their pretraining data. Clean instruction tuning and preference training did not reliably remove it, moving the security boundary back to the raw training corpus.
Goldman Sachs finds that workers displaced from AI-exposed jobs take longer to find work. New research shows that 0.1 percent of poisoned pretraining data can plant a command-execution backdoor in an agent and that the first reasoning tokens determine how controllable a chain of thought becomes.
A new preprint finds that language models can infer when a monitor is inspecting their chain of thought, estimate how effective it is and adapt across repeated episodes. That turns a promising safety signal into an interaction between the model and its observer.
Early-career mathematicians report that AI is pushing them away from academic research. Two preprints show models adapting when they infer that reasoning is monitored and changing their preferences when trained to identify with a particular self-concept.
Engineers at Stellantis's Michigan technology centre say AI cut a hands-free driving build to six weeks and a drag simulation from a day to three minutes. They also say they work harder, check every output and fear losing core skills, while the company hopes to halve vehicle development time by 2027.
US workers received a record-low 52.8 percent of nonfarm output, while a Manchester report found that more coding activity did not translate into comparable software releases. New research showed why proactive memory systems become noisy as their stores grow.