Radar / AI security / Vijil DART
Vijil DART (Diamond Adaptive Red Teaming for Agents)
Automated red-teaming system that deploys its own adversarial agents to probe enterprise AI agents with adaptive multi-turn attacks, claiming up to 1.5x more security/safety issues found than static-prompt tools.
Why it matters
Red-teaming moves inside the agent development lifecycle — attacks that learn, aimed at the coming fleet of 150K+ enterprise agents.
What you could build with it
A platform team shipping customer-facing agents could run DART in CI on every agent release, turning adaptive attack findings into a regression suite that blocks deployment of agents that leak data under multi-turn probing.
Does it hold up?
Too early to judge — announced Oct 7; no independent usage evidence exists yet. Only evidence is the vendor-run DecodingTrust-Agent benchmark (1.5x vs closest competitor). The vijil CLI docs show a working red-team workflow (waves, seeds, judgments, reports), but DART-specific third-party builds will take weeks to appear.
Built with Vijil DART
- vijilai/vijil-travel-agentgithub · Enterprise travel-agent demo repo for Vijil's Diamond/Dome/Darwin platform, including seeded vulnerabilities used in red-team evaluations (verified via GitHub API).
Learn more
First spotted on article: source.
More AI security
OpenAI, Anthropic and Google DeepMind jointly unveil cyber-focused safety models and safeguardsThe three rival labs disclosed weeks of behind-the-scenes coordination and jointly released a set of cyber-focused AI…security · JEV 0.73OpenAI textGrain watermarking for ChatGPT and Codex in the EUOpenAI will roll out its invisible textGrain watermarking system to ChatGPT/Codex users in the EU over the coming…security · JEV 0.65Cloudflare security-audit-skill: multi-phase security audits for coding agentsA coding-agent skill that runs multi-phase security audits with independently verified, machine-checkable findings.security · JEV 0.63LiveNerf: a pre-registered benchmark for post-release model driftOpen-source, pre-registered 30-day benchmark that detects whether Claude Opus 5.5 quietly gets worse after launch.security · JEV 0.61Sandlock 0.8.8: deferred commit for agent sandboxesThe process-based Linux AI sandbox (no container, no VM) ships deferred commit: every run returns a changeset of what…security · JEV 0.58OpenAI disrupts Moonshot-linked 'adversarial distillation' campaignOpenAI disclosed it shut down a coordinated campaign by 15,000+ users to extract hidden model reasoning, attributing a…security · JEV 0.58
Get the week's best AI launches, plus 3 ideas worth building
One email every Saturday. Ranked by traction, not hype. Free.