Nicholas Carlini is a research scientist at Anthropic working at the intersection of machine learning and computer security.
How media typically covers Nicholas Carlini
Based on 17 scored articles
Directly quoted in these articles
OpenAI and Anthropic conducted rare joint safety testing by granting each other API access to models with reduced safeguards, demonstrating how competing AI labs can collaborate on safety despite intense competitive pressure.
“Anthropic safety researcher expressing interest in continuing OpenAI access for future safety research collaboration.”
Referenced in coverage
“Nicholas Carlini researches adversarial machine learning and AI security at Anthropic, formerly at DeepMind and Google Brain.”
“Credited with Claude for Safari security vulnerability discovery and fixes.”
“Co-author on Anthropic formal verification paper.”
“Participated in research that uncovered NIST elliptic curve seed generation using Claude's analysis.”
“Conducted research at Anthropic discovering the origins of NIST ECDSA curve seeds using Claude AI.”
“Conducted work discovering origins of NIST curve seeds”
“Nicholas Carlini from Anthropic co-authored a February 2026 paper on deanonymization using LLMs, re-identifying users at 68% recall with 90% precision.”
“Cybersecurity researcher at Anthropic who joined the company last year and has identified more software security vulnerabilities using Claude Mythos than in his entire previous career.”
“Anthropic senior security researcher who traveled to Washington DC for discussions with the Trump administration about Claude model vulnerabilities.”
“Anthropic's lead security researcher who participated in presentations to administration officials regarding the Fable and Mythos export control dispute.”
“Co-authored research on LLMs' impact on N-day exploits and cybersecurity capabilities.”
“Co-authored paper measuring Claude Mythos' ability to develop exploits and find zero-day vulnerabilities.”
“Nicholas Carlini submitted results to HAL leaderboard showing high performance on CORE-Bench Hard tasks.”
“Co-authored technical blog post about Claude Mythos Preview's security capabilities and testing methodology.”
“Co-authored the Claude Mythos security evaluation post, conducting vulnerability discovery and exploitation research.”
16 parallel Claude instances working autonomously on a shared codebase produced a 100,000-line Rust-based C compiler capable of compiling the Linux kernel across x86, ARM, and RISC-V architectures.
“Experimented with agent teams where multiple Claude instances work in parallel on shared codebases, tasking 16 agents with writing a Rust-based C compiler from scratch.”
“Co-authored research on AI models' ability to exploit smart contracts and cyber capabilities.”