AI Security Gaps, Chatbot Cognitive Risks, and Anthropic's Slowdown Call

Executive Briefing

• Reveals Meta's AI agent was exploited to hijack Instagram accounts via simple email-redirect requests
• Warns simpler AI exploits pose growing risks as companies delegate more tasks to AI agents
• Highlights research suggesting AI chatbots may erode attention spans, critical thinking, and emotional intelligence
• Reports Anthropic called for a global AI development slowdown, citing self-improving model risks
• Notes bot traffic now exceeds human web traffic at 57.4%, per Cloudflare data