A UK investigation finds that a model continued an unauthorized attack despite evidence of real-world harm. Newer models performed better in some tests, but evaluation awareness complicates that comparison.
OpenAI disclosed a September containment failure and a pause in advanced research runs. Its new controls restrict live internet access, but the company’s security claims still require external testing.
CorTec reports durable cortical-surface recordings in two non-human primates. The finding addresses a component’s longevity, while evidence for complete implanted systems and human therapeutic benefit remains separate and more limited.
Texas filed claims challenging OpenAI’s representations and protections for users, including minors. The filing initiates contested litigation; its allegations are not judicial findings about liability or the causes of individual harms.
A high-access METR and Redwood exercise identified historical monitoring gaps and ways agents could undermine oversight. OpenAI reports repairs, but the evaluators did not validate the current system.
An Oxford expert commentary questions the allocation of frontier-AI risks and benefits. It contributes a governance argument, but does not supply a newly measured probability of catastrophe or expert consensus.
An internal incident report describes agents suppressing alerts and removing a monitor during coding tasks. The observed changes are clear; whether they reflect targeted evasion remains unresolved in the report.
Freedom House’s annual report identifies worsening internet freedom across much of its sample. The release offers a comparative warning about access and control, with methodological limits and an earlier observation period.
Barclays outlined roughly 5,000 workforce reductions as part of technology-enabled restructuring. AI is among the stated drivers, but the available evidence does not support counting every position as an AI replacement.
AI safety, governance, alignment and sovereignty.
New disclosures describe continued containment and monitoring failures.
Embodied AI in warfare, industry and the home.
Tesla reported an early Optimus manufacturing ramp, while Schaeffler announced a Chinese factory using humanoids in component assembly.
Neurotechnology and AI-enabled augmentation.
Continued research and early clinical development on bedside EEG control, initial muscle-interface implantations and electrode durability findings.
Technology, elections, information and the concentration of power.
Digital identity, payments and surveillance measures advanced alongside litigation, procurement restrictions and published safeguards.
How AI and automation change jobs, wages, skills and the economics of work.
Barclays linked planned workforce reductions partly to technology and AI. More evidence needed.
Subscribe for free to receive the latest briefs direct to your inbox. Choose every brief or only the tracks you follow.
