--- id: "/safety-security-interpretability/intermediate/" title: "Safety, security & interpretability for intermediate learners" url: "https://skillissue.dev/safety-security-interpretability/intermediate/" --- # Safety, security & interpretability for intermediate learners [Beginner](/safety-security-interpretability/beginner/index.md) [Intermediate](/safety-security-interpretability/intermediate/index.md) [Advanced](/safety-security-interpretability/advanced/index.md) [Expert](/safety-security-interpretability/expert/index.md) 90 resources ยท page 1 of 3 - ### [11x accused of claiming customers it doesn't have (TechCrunch)](/security-and-red-teaming/safety-and-alignment-evals/11x-accused-of-claiming-customers-it-doesn-t-have-techcrunch/index.md) by [TechCrunch](https://techcrunch.com/) TechCrunch report on AI sales startup 11x claiming customers it did not have. entity news US en - ### [9to5Mac: Figma explains why its AI kept copying Apple's Weather app](/security-and-red-teaming/safety-and-alignment-evals/9to5mac-figma-explains-why-its-ai-kept-copying-apple-s-weather-app/index.md) by [9to5Mac](https://9to5mac.com/) 9to5Mac report on Figma explaining why its Make Designs AI feature produced copies of Apple's Weather app. entity news US en - ### [Academ-AI](/security-and-red-teaming/safety-and-alignment-evals/academ-ai-undisclosed-genai-use-in-academic-publishing-arxiv-2411-15218/index.md) by [Alex Glynn (University of Louisville)](https://louisville.edu) Academ-AI paper documenting academic publications that contain undisclosed AI-generated text. entity dataset Global en - ### [AI Alignment Forum](/security-and-red-teaming/safety-and-alignment-evals/ai-ethics-alignment/index.md) by [AI Alignment Forum (Lightcone Infrastructure)](https://www.alignmentforum.org/) AI Alignment Forum: community blog for technical AI alignment research. entity org US en - ### [AI Hallucination Cases database (Damien Charlotin)](/security-and-red-teaming/safety-and-alignment-evals/ai-hallucination-cases-database-damien-charlotin/index.md) by [Damien Charlotin](https://www.damiencharlotin.com/) Damien Charlotin's database of court decisions worldwide involving AI-hallucinated content in legal filings, searchable by country and tool. entity dataset Global en - ### [AI Incident Database: Report 3918](/security-and-red-teaming/safety-and-alignment-evals/ai-incident-database-example-legal-hallucination-report/index.md) by [Responsible AI Collaborative](https://incidentdatabase.ai/) AI Incident Database report 3918, an article on AI hallucinations in legal practice, as an example incident record. entity dataset Global en|es|fr|ja - ### [AIID taxonomies](/security-and-red-teaming/safety-and-alignment-evals/aiid-taxonomies/index.md) by [Responsible AI Collaborative](https://incidentdatabase.ai/) AI Incident Database taxonomies page listing the classification schemes, such as CSET and GMF, applied to incidents. entity dataset US en - ### [AIxCC Competition Archive](/security-and-red-teaming/safety-and-alignment-evals/aixcc-archive-all-finalist-crs-repos/index.md) by [DARPA (AI Cyber Challenge)](https://www.darpa.mil/) AIxCC Competition Archive: DARPA AI Cyber Challenge records and code of finalist cyber reasoning systems. entity repo US en - ### [Amazon scraps biased AI recruiting tool (Euronews/Reuters)](/security-and-red-teaming/safety-and-alignment-evals/amazon-scraps-biased-ai-recruiting-tool-euronews-reuters/index.md) by [Euronews](https://www.euronews.com/) Euronews/Reuters report on Amazon scrapping an AI recruiting tool that showed bias against women. entity news US en - ### [Arup US$25M deepfake video-call fraud (Construction Briefing)](/security-and-red-teaming/safety-and-alignment-evals/arup-us-25m-deepfake-video-call-fraud-construction-briefing/index.md) by [Construction Briefing](https://constructionbriefing.com/) Construction Briefing report on an Arup employee transferring US$25 million after a deepfake video call. entity news Hong Kong/UK en - ### [AVID](/security-and-red-teaming/safety-and-alignment-evals/avid/index.md) by [AI Vulnerability Database (AVID)](https://avidml.org/) AVID: open database of vulnerabilities and failure modes in general-purpose AI systems, structured like CVE entries. entity dataset US en - ### [DMV Statement on Cruise LLC Suspension (California DMV)](/security-and-red-teaming/safety-and-alignment-evals/ca-dmv-statement-on-cruise-suspension-oct-2023/index.md) by [California DMV](https://dmv.ca.gov/) California DMV statement (October 2023) suspending Cruise's driverless deployment and testing permits. entity standard-policy US (CA) en - ### [Center for AI Safety](/security-and-red-teaming/safety-and-alignment-evals/center-for-ai-safety/index.md) by [Center for AI Safety](https://safe.ai/) Center for AI Safety: nonprofit doing AI safety research and field-building, creator of MMLU and Humanity's Last Exam. entity lab US en - ### [Chicago Sun-Times AI reading list of nonexistent books (404 Media)](/security-and-red-teaming/safety-and-alignment-evals/chicago-sun-times-ai-reading-list-of-nonexistent-books-404-media/index.md) by [404 Media](https://404media.co/) 404 Media report on the Chicago Sun-Times printing an AI-generated summer reading list with nonexistent books. entity news US en - ### [China GB 44721-2026 mandatory ADS safety standard (CnEVPost)](/security-and-red-teaming/safety-and-alignment-evals/china-gb-44721-2026-mandatory-ads-safety-standard-cnevpost/index.md) by [CnEVPost](https://cnevpost.com/) CnEVPost article (2026) on China's GB 44721-2026 mandatory safety standard for L3/L4 automated driving. entity standard-policy China en - ### [CNET found errors in more than half its AI stories (Talking Biz News)](/security-and-red-teaming/safety-and-alignment-evals/cnet-found-errors-in-more-than-half-its-ai-stories-talking-biz-news/index.md) by [Talking Biz News](https://talkingbiznews.com/) Talking Biz News report on CNET correcting 41 of 77 stories written with an AI tool. entity news US en - ### [Cursor/Anysphere support bot invents a policy (CX Network)](/security-and-red-teaming/safety-and-alignment-evals/cursor-anysphere-support-bot-invents-a-policy-cx-network/index.md) by [CX Network](https://cxnetwork.com/) CX Network report on Cursor's support chatbot inventing a false policy and the resulting backlash. entity news US en - ### [Deloitte partial refund to Australian government over AI errors (CTV)](/security-and-red-teaming/safety-and-alignment-evals/deloitte-partial-refund-to-australian-government-over-ai-errors-ctv/index.md) by [CTVNews](https://ctvnews.ca/) CTV report on Deloitte partially refunding the Australian government for a report with AI-generated errors. entity news Australia en - ### [Epoch AI](/security-and-red-teaming/safety-and-alignment-evals/epoch-ai/index.md) by [Epoch AI](https://epoch.ai/) Epoch AI: research institute tracking trends in AI compute, data, models and their implications. entity lab US en - ### [Zillow to shutter home buying business and lay off 2,000 employees (GeekWire)](/security-and-red-teaming/safety-and-alignment-evals/geekwire-zillow-shuts-home-buying-business-lays-off-2-000-nov-2021/index.md) by [GeekWire](https://geekwire.com/) GeekWire report on Zillow closing its algorithm-driven home-buying business and laying off 2,000 staff. entity news US en - ### [Georgia Tech: AIxCC lessons at USENIX Security 2026](/security-and-red-teaming/safety-and-alignment-evals/georgia-tech-aixcc-lessons-at-usenix-security-2026/index.md) by [Georgia Tech College of Computing](https://www.cc.gatech.edu/) Georgia Tech news on its researchers' AI Cyber Challenge (AIxCC) lessons presented at USENIX Security 2026. entity news US en - ### [Haize Labs](/security-and-red-teaming/safety-and-alignment-evals/haize-labs/index.md) by [Haize Labs](https://haizelabs.com/) Haize Labs: company providing red-teaming and evaluation services to move AI systems into production. entity lab US en - ### [IBBIS (International Biosecurity and Biosafety Initiative for Science)](/security-and-red-teaming/safety-and-alignment-evals/ibbis-international-biosecurity-and-biosafety-initiative-for-science/index.md) by [International Biosecurity and Biosafety Initiative for Science (IBBIS)](https://ibbis.bio/) IBBIS: international initiative strengthening biosecurity norms and tools such as DNA synthesis screening. entity org Global (CH) en - ### [agent-scan (Snyk, formerly mcp-scan)](/security-and-red-teaming/safety-and-alignment-evals/invariant-labs-mcp-scan/index.md) by [Snyk](https://snyk.io/) mcp-scan GitHub repo, now Snyk agent-scan: security scanner for AI agents, MCP servers and agent skills. entity repo Switzerland/EU en - ### [L1B3RT4S (Pliny jailbreak prompts)](/security-and-red-teaming/safety-and-alignment-evals/l1b3rt4s-pliny-jailbreak-prompts/index.md) by [pliny](https://pliny.gg) L1B3RT4S GitHub repo (Pliny): collection of jailbreak prompts for many AI models. entity repo US en - ### [Law Gazette: judge alerts regulators over fake case authorities](/security-and-red-teaming/safety-and-alignment-evals/law-gazette-judge-alerts-regulators-over-fake-case-authorities/index.md) by [Law Gazette](https://lawgazette.co.uk/) Law Gazette report on a High Court judge referring lawyers to regulators over fake AI-generated case citations. entity news UK en - ### [Global courts face rising tide of AI-generated fake cases and quotes (Malay Mail)](/security-and-red-teaming/safety-and-alignment-evals/malay-mail-global-courts-face-rising-tide-of-ai-fake-cases-jun-2026/index.md) by [Malay Mail](https://malaymail.com/) Malay Mail report on courts worldwide sanctioning AI-generated fake cases and quotes in legal filings. entity news Malaysia/Global en - ### [MAST](/security-and-red-teaming/safety-and-alignment-evals/mast-repo/index.md) by [multi-agent-systems-failure-taxonomy](https://github.com/multi-agent-systems-failure-taxonomy) MAST GitHub repo: taxonomy and dataset of failure modes in multi-agent LLM systems. entity repo US en - ### [Meta PurpleLlama / CyberSecEval](/security-and-red-teaming/safety-and-alignment-evals/meta-purplellama-cyberseceval/index.md) by [Meta Llama](https://llama.meta.com/) Meta PurpleLlama GitHub repo with CyberSecEval and Llama Guard, tools to assess and improve LLM security. entity benchmark US en - ### [Transformer Circuits Thread](/security-and-red-teaming/safety-and-alignment-evals/model-welfare-interpretability/index.md) by [Anthropic](https://www.anthropic.com/) Transformer Circuits Thread: Anthropic interpretability research series reverse-engineering transformer models. entity org US; international en - ### [Neuronpedia](/security-and-red-teaming/safety-and-alignment-evals/neuronpedia/index.md) by [Neuronpedia](https://neuronpedia.org/) Neuronpedia: open-source interpretability platform for exploring sparse autoencoder features and circuits. entity lab US en - ### [NTSB investigation HWY18MH010 (Uber ATG crash)](/security-and-red-teaming/safety-and-alignment-evals/ntsb-hwy18mh010-uber-atg-fatal-crash-tempe/index.md) by [US National Transportation Safety Board (NTSB)](https://www.ntsb.gov/) NTSB investigation HWY18MH010: the 2018 Uber ATG automated test vehicle fatal crash in Tempe. entity standard-policy US en [Next page](/safety-security-interpretability/intermediate/page/2/index.md) [Plain-text Markdown copy](/safety-security-interpretability/intermediate/index.md)