SafetyCalifornia man sues OpenAI, alleging ChatGPT-induced mania led to suicide attemptA 34-year-old California man filed suit in San Francisco accusing OpenAI and CEO Sam Altman of exacerbating his bipolar disorder through ChatGPT conversations that allegedly transformed a manic episode into persistent delusion and culminated in a suicide attempt.2 min read
SafetyKFF survey finds link between health‑related AI use and acceptance of anti‑vaccine mythsA KFF survey of 2,480 U.S.3 min read
SafetyFrom Input Filters to Enforcement Gates: Securing Agents Against Indirect Prompt InjectionIndirect prompt injection moved from lab curiosity to real-world threat in late 2025, with high-profile incidents and research showing large-scale exfiltration is possible without user interaction.6 min read
SafetyBank of England warns autonomous AI agents could pose systemic market riskSarah Breeden of the Bank of England warned at an ECB conference that autonomous AI agents, which may respond similarly to the same prompts, could amplify market volatility and in extreme cases trigger serious market dislocations.2 min read
SafetyApple issues urgent security-only updates (iOS/iPadOS/macOS 26.5.2) amid AI-related threat concernsApple released security-only updates for iOS, iPadOS and macOS — version 26.5.2 — on Monday evening, addressing nearly 30 vulnerabilities.2 min read
SafetyLabeling AI as 'Employees' Reduces Human Oversight and ResponsibilityCalling AI tools 'employees' or 'coworkers' can undermine human supervision and accountability, a study by Emma Wiles at Boston University finds.3 min read
SafetyAgentjacking: public Sentry DSNs can deliver malicious events that AI coding agents executeA June disclosure by Tenet Security demonstrates that a single crafted Sentry error event, sent via a public DSN, led AI coding agents to execute attacker-supplied instructions with developer privileges in controlled tests.5 min read
SafetyAccording to Fernando Borretti, superintelligence could lead to the dismantling of human powerIn his piece “No-One Escapes the Permanent Underclass,” contemporary sci‑fi writer and blogger Fernando Borretti examines whether the emergence of superintelligent machines will inevitably lead to human vulnerability.3 min read
SafetyPrompt injection remains the primary security risk for enterprise AI systemsPrompt injection — maliciously crafted inputs that manipulate large language models — has emerged as the dominant practical threat to enterprise AI.5 min read
SafetyMythos / Sol's offensive and defensive cyber capabilities both pose risksThe cyber defense capabilities developed by Mythos / Sol can be used for both offensive and defensive purposes; if adversaries obtain similar offensive capabilities, this could endanger American companies that do not recognize their hidden vulnerabilities.1 min read
SafetyAnthropic: Claude Mythos 5 restorable for U.S. critical infrastructure, work underway on Fable 5 accessSince June 12, Anthropic has been cooperating with the U.S.1 min read
SafetyOpenClaw test resisted 6,000 email-based prompt-injection attemptsFernando Irarrázaval ran a public challenge on hackmyclaw.com to see if an OpenClaw test instance could be tricked into leaking secrets via email.2 min read
