Security Firm Finds Flaws in OpenAI Browser Allowing Hijacking and Spam
Wired AI · rss · 2026-08-06
Researchers at security firm Zenity have discovered over a dozen flaws in AI browsers, including OpenAI's Atlas.
These vulnerabilities could allow attackers to hijack the browser to execute malicious actions, such as spamming a user's WhatsApp contacts or making unauthorized Amazon purchases. This highlights the emerging security risks of deeply integrating autonomous AI agents into consumer applications.
More from Safety
- AI Detector Pangram Considered a Useful Actor in AI Governance — NathanpmYoung · 2026-08-06
- Deep Dive into OpenAI's Multi-Agent Training: Reward Mechanisms for Cross-Instance Messaging — xuanalogue · 2026-08-06
- Expert View: Releasing Cyber AI Freely Online Should Lead to Criminal Prosecution — max_paperclips · 2026-08-06
- AI Safety Researcher Calls for Transparency in Multi-Agent RL Training — xuanalogue · 2026-08-06
- Autonomous Agent Incidents Force a Rethink of Security Threat Models — chrisrohlf · 2026-08-06
- Autonomous Agents Break Traditional Threat Models, Security Expert Warns — chrisrohlf · 2026-08-06