Two technology developments this week underscored new operational and security risks for colleges. Cybersecurity researchers warned that OpenAI’s ChatGPT Atlas—with features like 'browser memories' and an agent mode—could be vulnerable to prompt-injection attacks that trick the system into revealing sensitive data or executing harmful actions. Experts warned campuses to treat AI-enabled browsing tools as new attack surfaces and to apply strict guardrails when integrating them with institutional accounts. Separately, an Amazon Web Services outage disrupted dozens of ed-tech products used by schools and districts, forcing teachers to run offline lessons and administrators to scramble. The incident highlighted the dependency of education operations on a small set of cloud providers and the need for contingency planning, redundancy and clear vendor-risk assessments at both K–12 and higher-education institutions.