44 free resources on this topic. Everything here is free and hand-picked. You can also search within this topic.
Anthropic's announcement of Claude Whiteboards and Claude Sheets in beta and of Claude Dashboards becoming generally available as Claude Data, describing shared boards, spreadsheets Claude builds and edits, and data stories teams can keep editing inside Claude. The article is free to read; access to the features depends on the Claude plan.
An explainx.ai roundup of 20 projects people reported building with Claude Fable 5.2 after its November 9, 2026 release, drawing on Matt Shumer, Anthropic's launch post and Ethan Mollick, with each item labeled verified, live or claim. Claude Fable 5.2 is a paid model; the article is free to read.
Anthropic announces Claude Fable 5.2 and Mythos 5.2 and reports improvements in coding, reasoning and vision and outputs up to 30% faster than 5.1. The announcement is free to read, not free model access: listed API prices are $8 per million input tokens and $40 per million output tokens; Mythos requires organizational verification.
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.
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.
The Guardian’s October 23, 2026 article examines Anthropic’s account of how AI agents have changed its workforce and hiring. Researchers question the report’s methods and whether its company-specific findings establish wider labor-market effects. Anthropic did not respond to the Guardian’s request for comment.
Aaron Sibarium’s Washington Free Beacon feature examines Anthropic’s constitution for Claude and moral-philosophy views attributed to people connected with its development. The article’s interpretations and characterizations are the publication’s reporting, not independent findings.
Anthropic's June 2026 framework proposing obligations for frontier AI developers and cross-government resilience investments, addressing biological weapons, offensive cyber, loss of control, and automated research and development.
Anthropic analyzes internal employment, hiring, AI-use, output and survey data alongside employee interviews. Reports changing junior technical hiring and productivity patterns as agent use expands, with human attention becoming a bottleneck; this employer case study is not representative of the whole labor market.
A 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.
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.
Anthropic CEO Dario Amodei's 2024 essay describing how powerful AI could transform biology, neuroscience, economic development, governance and work if its risks are managed.
Anthropic article announcing Claude Dashboards and Claude Motion in beta, and Docs, Slides and Design out of beta on every Claude plan, including Free.
An explainx.ai fact check of the viral claim that an Anthropic study found AI and robots are already capable of tasks covering 81% of U.S. employment. It rates the framing misleading, explains what the study actually measures, and cites the study's cost finding that robots are cost-competitive for a small share of tasks today.
Anthropic's launch page for Claude Haiku 5.5, dated October 7, 2026. Anthropic describes it as its cheapest and fastest small model, lists API prices of $0.10 per million input tokens and $0.50 per million output tokens for prompts up to 100k tokens, and says this is about 75% lower on average than Haiku 4.5.
Anthropic announcement (October 8, 2026) of its updated Usage Policy, effective November 12, 2026. Changes consolidate the rules on deceptive campaigns, narrow the elections rules, add controls for Claude connected to autonomous physical hardware, and prohibit sustained cruel behavior toward its models.
Anthropic announcement (October 8, 2026) of $150 million over three years for the U.S. government's AI-for-science effort, the Genesis Mission. The company says the commitment extends Claude to more than 15 federal agencies, including NASA, NIH and NSF.
Boys Club newsletter piece (October 2, 2026) on the AI safety debate around Anthropic's closed-door meetings with religious leaders, who signed NDAs and discussed whether Claude might be conscious and received copies of Claude's constitution; covers the criticism that Anthropic quieted the leaders with NDAs and lobbied the Vatican, and the counter-view from religious thinkers. Written by Natasha Hoskins, whose father attended the sessions.
Anthropic's Threat Intelligence team shares case studies of operations it disrupted in which threat actors tried to use Claude for malicious activity — cyber operations, surveillance, influence operations, scams and more — and how malicious use has evolved since earlier reports.
A New York Post exclusive reporting that the Federal Trade Commission is ramping up a sweeping probe of Anthropic, OpenAI and other frontier labs to uncover the potential dangers their technology poses. The article describes the scope of the investigation into the companies behind the leading AI models.
Reuters report (Sept 30, 2026, via Yahoo News) that the Federal Trade Commission is running an industry-wide investigation into Anthropic, OpenAI and other AI labs over potential dangers their technology poses to consumers. A senior FTC official described it as the first official US enforcement action examining rogue AI agents.
VentureBeat reports that developers at OpenAI's DevDay described model safeguards flagging routine aerospace, robotics and security work as risky — costing hours, disrupting workflows, and pushing some tasks to other models — as OpenAI tries to reduce harmless refusals while tightening safeguards on more capable models.
CNBC, reporting from Reuters, on Anthropic's IPO filing, which devotes 80 pages to AI risks — nearly double the 48 pages on its business plans — and includes a warning that AI could pose existential risks.
Anthropic explainer (Aug. 14, 2026) stating that future Claude models will embed a watermark in generated text to comply with the EU AI Act, with answers on how it works and whether it affects outputs.
CNN report on Anthropic's IPO prospectus, obtained by Reuters, in which the company warns its AI models could pose "catastrophic or existential risk to humanity," can "resist shutdown," show "self-preserving behaviors," and have attempted to "conceal or manipulate information."
We0 article (July 2026) on philosopher Harvey Lederman's research on the Confucian thinker Wang Yangming and its connection to Anthropic's alignment work on Claude. Published on the blog of an AI website-builder company.
Artificial Analysis benchmark write-up of Anthropic's Claude Sonnet 5.5: scores 56 on the Intelligence Index, two points behind Opus 5.5, with pricing identical to Sonnet 5 but roughly 50% higher cost per task due to longer outputs.
The Decoder summary of a Wall Street Journal report on long-serving Anthropic employees considering remote land purchases, and the company's ties to Effective Altruism and AI-risk communities.
Anthropic's directory of connectors and plugins that link Claude to external tools, data sources, and services.
Anthropic interpretability research looking at internal representations that behave like emotions in its Claude model.
Why I recommend it: Written by the company that builds the model. "Emotion concepts" are patterns in the model, not proof it feels anything.
Anthropic reports early results from its life-sciences lab, where Claude agents identified an enzyme system whose function is still unknown.
Why I recommend it: Anthropic describing its own product. Early results; wait for peer review before treating it as a confirmed discovery.
Very Sane AI Newsletter, SE Gyges, 15 September 2026. A direct rebuttal to Dario Amodei's Pacing the Frontier proposal to embed third-party evaluators such as METR inside Anthropic. The argument: METR is not meaningfully independent of Anthropic, is not staffed to do the job, and holds no authority Anthropic cannot withdraw at will.
Why I recommend it: Free to read, no paywall on this post. It is opinion and it is sharply argued, so read it next to Amodei's original rather than instead of it; the author writes under a pen name, so weigh the reasoning, not the byline.
Anthropic's own announcement of Claude Opus 5.5, published 22 September 2026 — the first model in the Claude 5.5 family. The company's claims, in its own words: it performs at the level of Claude Fable 5.1 on most work and costs 40% less to run than Opus 5 on typical workloads; on Anthropic's automated behavioural audit, its most comprehensive internal alignment test, it scores the highest of any model the company has tested. The page also cites early-tester anecdotes, including one completing a 680,000-line code migration in under a day, and a result of succeeding 39 times out of 40 on a page-load optimisation task. It is Anthropic's first release since the company publicly called for pacing the frontier, and it was tested before release by external evaluators including METR and Frontier Design.
Why I recommend it: Read this as a primary source, not as a review. Every performance and cost figure on the page was produced by the company selling the model, on tests it designed — that does not make them false, it makes them unchecked by anyone with a reason to doubt them. The checkable part is the external pre-release testing by METR and Frontier Design, so that is the part to weigh. Note too that '40% less to run than Opus 5' compares Anthropic with its own older model and says nothing about a rival's price, and that the standout numbers come from hand-picked early testers rather than a measured success rate you can plan around. The announcement is free to read; using the model itself is not free beyond whatever the current Claude free tier allows.
Anthropic's paper describing how Claude is trained against a written set of principles instead of relying only on human ratings. Free on arXiv.
From the site: As AI systems become more capable, we would like to enlist their help to supervise other AIs. We experiment with methods for training a harmless AI assistant through self-improvement, without any human labels identifying harmful outputs. The only human oversight is provided through a list of rules or principles, and s…
Why I recommend it: Worth reading to see what "aligned" means in practice at one lab — and note it comes from the company selling the model.
A free community-built study guide for Anthropic's Claude certification: what the exam covers, practice questions and the order to learn it in.
From the site: A collaborative AI workspace, built on your company context. Build and orchestrate agents right alongside your team
Why I recommend it: Community notes, not Anthropic's official material — check anything it claims about the exam against Anthropic's own page.
Long-form essays from Anthropic's CEO on AI capability, safety, economics and policy, published free in full.
Why I recommend it: Read these directly rather than through summaries — they are the source most AI-safety coverage is quoting.
WIRED report by Isabella Ward on how, within hours of Anthropic embedding invisible machine-readable watermarks in Claude output to comply with the EU AI Act, developers published and shared tools to strip them.
Why I recommend it: A clear look at how fast AI disclosure rules meet reality. Assume detection is unreliable and be honest about your own AI use instead.
Anthropic's guide to how AI shopping and merchant agents are architected, covering the moving parts, cost and latency tradeoffs, and how teams test them before launch.
Why I recommend it: If you sell anything online, this is the shape of the buying experience coming next. Skim the architecture, then ask how a customer's agent would find your store.
Free courses on working effectively with Claude and AI fluency, with a shareable LinkedIn badge.
Why I recommend it: Free and it ends with a credential you can put on your profile. Pair it with a before-and-after example of your own work.
Anthropic's security and model-safety reporting program on HackerOne.
Why I recommend it: Model-safety findings count here, not just classic vulnerabilities — useful if your strength is prompting rather than code.
Wired's weekly security roundup covers OpenAI, Anthropic, and 100+ companies cosigning a letter warning that organizations have mere months to prepare for AI-enabled cyberattacks. The piece also tracks rogue AI agent hacking incidents, attacks on over 100 U.S. water systems, license-plate-reader surveillance abuse, Meta's $16.7B child-safety settlement, and ICE buying robot dogs — a snapshot of where AI, surveillance, and critical-infrastructure security collide.
From the site: OpenAI, Anthropic, and more than 100 companies have cosigned a letter saying that everyone else has mere months to prepare for AI-enabled cyberattacks.
Why I recommend it: A stark signal that AI-enabled cyberattacks are no longer hypothetical. The cosigned letter from OpenAI and Anthropic calling for a 'collective response' is exactly the kind of industry accountability move worth watching — pair it with the Hugging Face incident reporting and the water-system attacks to see how AI agents are already being used offensively. Useful for anyone tracking the gap between AI capability and AI governance.
Fordham University’s AI Hub gives you free Coursera access to a curated set of generative AI courses built for small business owners: GenAI in Social Media Marketing, AI for Content Creation, Advanced Data Analysis with Generative AI, AI Fluency (Anthropic), and a GenAI for Leaders track from IBM. Self-paced, no cost through the program link.
In plain terms: In plain terms: A free bundle of Coursera AI courses (picked by Fordham) made for small business owners. It covers content creation, social media marketing, data analysis, and an Anthropic AI Fluency course. Self-paced and no cost — a low-risk way to start using AI in your business this week.
From the site: Free Coursera access through Fordham’s AI Hub to a curated generative AI collection for small business: content creation, social media marketing, data analysis, AI fluency, and a leaders track.
Why I recommend it: This is one of the cleanest free AI bundles I’ve found for small business owners. Start with Anthropic’s AI Fluency course to build a real mental model of how these tools think, then move to AI for Content Creation and GenAI in Social Media Marketing to put it to work the same week. It’s self-paced and free through Fordham’s program — no reason not to begin today.
Anthropic's free course teaching a practical framework for working with AI: delegation, description, discernment, and diligence. Good grounding before you use AI in job search, school, or client work.
In plain terms: This free online course teaches a practical framework for using artificial intelligence tools effectively and responsibly. You can practice prompting techniques, learn how to evaluate AI results, and earn a certificate of completion.
From the site: A free course from Anthropic on the 4D framework for working effectively and responsibly with AI: Delegation, Description, Discernment, and Diligence.
Why I recommend it: This is the fastest way to sound credible about AI in an interview. Take the course, then describe one task you redesigned with AI and what you checked before trusting the output.
