Anthropic Research
ID: b6040af0-1bcc-5815-a08f-b6ee8cf87625
STIX ID: identity--b6040af0-1bcc-5815-a08f-b6ee8cf87625
Feed Type: skeleton
Earliest post: 2023-11-03
Latest post: 2026-07-28
Anthropic’s research explores AI safety, model interpretability, economic impact, and emerging risks, with the goal of making increasingly capable AI systems helpful, honest, and safe.
All
01/01/2020
08/11/2026
| Title | Date Published ↓ | Describes Incident | Author | Visible | |||
|---|---|---|---|---|---|---|---|
 | Discovering cryptographic weaknesses with Claude | 2026-07-23 | True | True | |||
| Measuring LLMs’ ability to develop exploits | 2026-06-05 | True | True | |||
| Assessing Claude Mythos Preview’s cybersecurity capabilities | 2026-04-16 | True | True | |||
| AI agents find smart contract exploits | 2026-03-24 | True | True | |||
| Cyber toolkits for LLMs | 2026-03-24 | True | True | |||
| Reverse engineering Claude's CVE-2026-2796 exploit | 2026-03-24 | True | True | |||
| AI models on realistic cyber ranges | 2024-11-02 | True | True | |||
| Many-shot jailbreaking | 2024-03-29 | True | True | |||
| A Mathematical Framework for Transformer Circuits | 2023-12-18 | True | True | |||
| Measuring LLMs’ impact on N-day exploits | 2023-11-03 | True | True | |||
| A small number of samples can poison LLMs of any size | 2023-11-03 | True | True | |||
| Mapping AI-enabled cyber threats | 2023-11-03 | True | True | |||
| Project Glasswing: An initial update | 2023-11-03 | True | True | |||
| Mitigating the risk of prompt injections in browser use | 2023-11-03 | True | True | |||
| LLM-discovered 0 days | 2023-11-03 | True | True |