Verified news, launches and analysis to understand what is changing in artificial intelligence and why it matters to you.
166 articles
The available documentation does not confirm the policy update described in the proposal, or its date, specific prohibitions, or enforcement mechanisms. Here is what can be distinguished from what still needs verification.
Anthropic has published a page titled “Anthropic Cyber Mission,” but the available information does not establish the initiative’s scope, participants, or safeguards. Its relationship to Project Glasswing and other cybersecurity programs also remains unclear.
Google’s documentation describes Gemini Enterprise as a workplace AI platform that can connect to applications and support knowledge work. However, the verified sources do not confirm that the agent has its own work identity or explain the scope of a launch involving autonomous delegation across apps.
Trust Layer adds four checks to a benchmark score: it examines the grader, the traceability of the answer, the truthfulness of the success claim, and consistency across runs. The framework was tested on five agent configurations and 108 tasks, but its results do not yet show that it works on other benchmarks.
OpenAI describes an experience that combines text with visual elements and interaction. The available information does not confirm which plans receive it or substantiate a 44% reduction in wait times.
Google’s documentation describes checking images and videos for SynthID watermarks. Detecting the watermark can provide a clue about a piece of content’s origin, but it cannot establish whether any file was created or edited with AI. The public launch, audio conditions and other details remain unconfirmed by the available information.
The proposal asks whether each retrieved memory is appropriate for a given query, but the verified material does not confirm the preprint’s existence, method or results. The distinction between a correct answer and authorized access remains a relevant question—not a finding that can be attributed to this work.
A preprint and project repository present JEPA-Anything as a framework for multiple domains. The available information does not confirm the tasks, experimental improvements, or scope of the reported preclinical result.
Cohere presents North 2 as an enterprise agent platform with controls for costs, security, and workflow management. The information available does not yet show how those controls work in production or which external models the platform supports.
The proposal aims to evaluate agents’ research processes, not just their final answers. But the available documentation does not confirm the preprint, its experiments, or its results.
The proposal describes a Critic for validating actions and a Context Manager for checking whether tasks are complete. Without a verifiable primary source, DeReAct’s publication and any results attributed to it on GAIA or SWE-bench Verified cannot be confirmed.
A preprint describes actions that violate safety constraints given much earlier. Its abstract reports an 11.5% rate of GHOST events in one GPT-5.5 setup and no observed events when testing STAR-Guard. Important details are still missing for judging how robust or broadly applicable those results are.