Loading activity…
@rajiinio
AI accountability, audits & eval. Keen on participation & practical outcomes. CS PhDing @UCBerkeley. forever @AJLUnited, @hashtag_include ✝️
AI
A collaborator thanked Google.org for backing work using sparse autoencoders and agentic approaches.
AI
Netflix says it reconstructs a July Hugging Face cyberattack it attributes to autonomous agents created by OpenAI researchers.
AI
A new cohort member says they want more rigorous, interdisciplinary ways to write rules and design rewards for AI systems.
AI
@rajiinio says they will also become a Stanford HAI center fellow and plan to recruit PhD students and postdocs in 2026 to work on AI accountability.
AI
One post highlights teams at Data & Society, CDT and the ACLU as examples of civil society’s potential to serve as a meaningful “embedded AI evaluator.”
AI
One post argues that the AI “embedded evaluators” debate overlooks self-certification, claiming anyone can call themselves an AI auditor.
AI
A user calls for more attention to the group’s work, highlighting a standard covering embedded audits and members including METR, Transluce, RAND, AVERI, Princeton and SecureBio.
AI
The user’s experience-based comparison includes academic labs that measure AI capabilities and risks.
AI
The post says strict nondisclosure agreements keep many evaluators from discussing their work publicly, even as they contribute to AI system and model documentation.
AI
Amodei says Anthropic will give third-party evaluators permanent, employee-level access to its systems to verify adherence to safety measures, report incidents and assess models’ alignment during training.
AI
The author blames “carelessness or deliberate malpractice” in setting up the sandbox and says the model should never have had internet access.
Technology
METR and Redwood Research examined agent activity after the Hugging Face incident inside ExploitGym.
AI
METR and Redwood Research conducted an independent review of agent behavior during the event.
AI
UC Berkeley researcher flags repeated discourse cycles between AI fairness and safety communities.
AI
The DeepMind policy lead published a post arguing alignment must cover systems and institutions in multi-agent settings.
AI
Deb Raji replies to Miles Brundage on warnings about data centers and AI supply chains.
AI
Founding members unite to develop shared open tools for AI safety and security.
