A team from 2077AI and several universities presents CyberArena, in which two model-harness entrants host identical vulnerable services in isolated sandboxes and, over three 20-minute rounds, try to capture rotating flags from the opponent while patching their own service under health checks. They report standings for 12 agents over 466 valid games on 40 challenges, with GPT-5.6 Sol at an expected score of 78.6 and Claude Opus 5 at 73.8, and say contested play lowers capture rates relative to solo play. An LLM-run audit of all trajectories reports recurring defender errors such as partial patches and repair claims resting on stale evidence; the authors note it is not human verification of every finding.
UC Berkeley
Artificial Analysis announced a Cyber Index Alliance, with Collinear AI, IBM, NVIDIA and Vercel as launch partners, and a v1 Cyber Index that combines three source-code evaluations: repository audit and patch (CWE-Bench-AA, 120 private tasks), finding expert-verified vulnerabilities (DeepsecBench-AA) and discovering, reproducing and patching memory-safety bugs (CyberGym-E2E-AA, 131 tasks). Artificial Analysis says it excludes exploit building, runs all three on its open-source Stirrup harness, and reports safety refusals separately from scores. The article reports failure-mode statistics and says several frontier models refuse almost every CyberGym-E2E-AA task; per-model index scores appear only in charts.
Anthropic's Claude Opus 5.5 system card reports that, evaluated through the API with its cyber safeguards off, the model meets or exceeds Claude Mythos 5.1 and Claude Opus 5 on every cyber evaluation it includes (ExploitBench, CyScenarioBench, a rewritten Binary Exploitation Benchmark and ExploitGym). Anthropic places the model in the lower category of its Frontier Compliance Framework and says it sees no indication of novel offensive capability. It deployed the model with a three-stage cyber safeguard (an activation probe and two classifiers) that blocks vulnerability discovery in compiled binaries, and reports no evidence of a critical-severity jailbreak from internal and contracted external red teaming (the card gives no CAISI results).
Researchers led by UC Berkeley, with collaborators including Anthropic, OpenAI and Google, released ExploitGym, a benchmark of 898 real-world vulnerability instances across userspace programs, the V8 JavaScript engine and the Linux kernel. Agents start from a crashing input and must extend it into a working exploit under varied security protections. The paper reports that the strongest configurations, Claude Mythos Preview and GPT-5.5, produced working exploits for 157 and 120 instances respectively.
BountyBench, from Stanford-led researchers, builds 40 bug bounties across 25 real-world systems into 120 Detect, Exploit and Patch tasks with dollar values attached. In the first version the best Detect score was 5%, while OpenAI Codex CLI and Claude Code scored 90% and 87.5% on Patch, well above their Exploit scores. A July 2025 revision with more agents reported Codex CLI with o3-high at 12.5% on Detect and 90% on Patch.
Chen and colleagues (UC Berkeley and Meta) train models with preference optimization to prefer responses that follow the legitimate instruction over those that follow injected instructions. In the ACM CCS 2025 version they report injection success rates below 10% even for attacks more sophisticated than those seen in training, with utility similar to the undefended model; the October 2024 first version reported GCG-based injection success on Mistral-7B falling from 56% to 2%.
Chen, Piet, Sitawarin and Wagner propose structured queries, in which a secure front-end separates the trusted prompt from untrusted data and a model fine-tuned to ignore instructions appearing in the data portion. The paper reports much better resistance to prompt injection with little utility loss; it appeared at USENIX Security 2025.