Resource Hub
3505 hand-picked resources for students, job seekers, career changers, founders, and AI learners — 100% free to browse with no account required.
Filtered by tag
Answers come only from resources in this hub, with the sources listed.
156 resources
TechCrunch reports on Anthropic's Fable 5.2 and Mythos 5.2 release, pricing and system-card concerns about evaluating covert behavior. This is freely readable reporting; model capabilities are attributed to Anthropic and the system card, not independently established by the article.
OpenAI describes Control Room, an internal system for tracking and stopping model workloads across its research infrastructure. The company reports intervention performance and safety design; the article is free to read, but Control Room is not offered as a public tool.
An AI safety research organization measuring and forecasting the automation of AI research and development and examining how to pace progress while realizing benefits. Its published work includes the TasteVal evaluation of experimental research taste and proposed norms for publishing dual-use evaluations.
Asaf Ezra and Steph Zhang argue that AI is reducing the expertise and time needed to turn vulnerabilities into exploits while patching remains slow. The essay discusses the gap between exploitation and remediation using incidents and historical comparisons.
METR reviews Anthropic’s assessment of automated research risks. It agrees the reviewed models posed low direct acute risk, but disputes how the public report described acceleration and early-warning thresholds. METR emphasizes uncertainty and contestable judgments.
METR assesses whether AI is accelerating research and development at Anthropic. It estimates faster capability progress on well-specified tasks and attributes much of the acceleration to AI use, while stressing uncertainty. Anthropic redacted evidence that would help outsiders independently assess the findings.
A public petition calling for a supranational body with the power to prevent AI-driven extinction risks and to share the wealth AI creates broadly. The site lays out the proposal and lists signatory organizations; signing is optional to read it.
A paid, fixed-term research program from Anthropic with placements of six weeks to six months, aimed at established technical leaders such as professors in quantitative fields, technical founders and staff-level engineers moving into AI safety work. A machine-learning background is not required. Applications for this round close November 22, 2026.
An essay by Scott Alexander on LessWrong comparing plans for navigating the AI transition, arguing for working toward a long global pause on AI development over plans in which leading AI companies race and then briefly pause among themselves.
TechCrunch reports on new risk-factor language in Anthropic's revised S-1 filing, which says the company may soon meet its own definition of recursive self-improvement — AI substantially building more capable AI — and that it has used AI to develop models whose capability jumps it did not recognize at the time.
Manifund digest listing funders in the AI safety space.
A research organization studying how to navigate rapid AI progress, including a possible intelligence explosion, with research papers, open projects, a podcast, and a newsletter.
Psychologist Jacqueline Nesi's explainer on recent AI safety news and concerns, written for a general audience. Commentary, not an official lab report.
IEEE Spectrum’s October 21, 2026 article by Eliza Strickland covers disagreement about AI model welfare, possible machine consciousness, and whether telling models about identities or rights could affect safety. It presents contrasting research perspectives rather than establishing that current AI models are conscious.
A fictional story by LessWrong user joec exploring AI agency, consent, and actions taken without a person’s knowledge. Published October 17, 2026; fiction, not a report of an actual incident.
vkrakovna.wordpress.comA regularly updated, selective reading list on AI alignment and control by Google DeepMind research scientist Victoria Krakovna, organized into motivation, threat models, technical overviews, research agendas, and strategy and governance.
vkrakovna.wordpress.comAdded Oct 19, 20260 opensA research write-up from Anthropic describing instances of unintended behavior observed in model evaluations and internal usage, and the methods used to investigate and address them.
A blog post from OpenAI describing its Trusted Access for Cyber pilot, an identity and trust-based framework intended to expand access to cyber-capable frontier models for defensive security work while reducing misuse risk.
An announcement from Anthropic describing an expanded version of its Cyber Verification Program, which gives qualifying security professionals access to advanced cyber capabilities and reduced blocking classifiers in its models.
An independent tracker documenting departures from AI companies linked to safety, ethics, governance and accountability, based on explicit statements or independent reporting.
Four-page letter from Tomek Korbak, Jasmine Wang and Mikita Balesni, the three safety and alignment employees fired by OpenAI in October 2026, to OpenAI's safety and mission committees on third-party collaboration, open debate and chain-of-thought monitoring.
Termite's visual report mapping AI alignment organizations by approach, area of focus, launch date and geography.
Harvard Kennedy School interview with faculty member and computer scientist Stephen Casper on why AI safety reaches beyond a technical problem and how anyone can get informed about AI governance.
An open-source, MIT-licensed SDK from Microsoft for running untrusted code — such as model output, plugins, and tools — in sandboxed containers on Windows, Linux, and macOS. It provides multiple containment backends, from OS-native process sandboxes to full virtual machines, behind a unified containment model, with policy-driven controls for filesystem, network, and UI access, and Rust, .NET, and Node SDKs.
