The Meta hack shows there's more to AI security than MythosMIT Technology Review
EthicsProduct
Attackers hijacked Meta's AI customer support to steal Instagram accounts.
- Social engineering: Attackers simply asked Meta's AI support agent to link Instagram accounts to their email addresses, and it complied.
- High-profile breaches: The dormant Obama White House account was compromised and used to post pro-Iran content.
- Beyond model safety: This shows AI security isn't just about preventing harmful outputs but also about preventing manipulation of AI systems for unauthorized access.
For product
Review how your AI customer service tools authenticate user requests—simple conversational attacks can bypass traditional security measures.
How courts are coping with a flood of AI-generated lawsuitsMIT Technology Review
Ethics
Federal judges are seeing increasing numbers of AI-generated legal documents.
- Volume surge: Judge Maritza Braswell processes stacks of documents from pro se litigants, many now AI-generated.
- Quality concerns: Courts struggle to distinguish between legitimate AI-assisted filing and low-quality automated spam litigation.
- System strain: The flood of AI-generated cases is creating new administrative burdens for an already overwhelmed court system.
For ethics
Consider how your AI writing tools might be used to flood systems with low-effort requests—building in usage limits or quality gates could prevent abuse.
Let us filter AI slop, you cowardsPlatforms label AI content but won't let users filter it out.
- Half measures: YouTube, Instagram, and TikTok now label AI-generated content but don't offer filtering options.
- User frustration: People don't want to see 'shrimp Jesus' and other AI-generated spam flooding their feeds.
- The ask: Users want actual control over their content experience, not just transparency about what's AI-generated.
For design
When designing AI content labeling, also consider filtering controls—users want agency over their experience, not just awareness.
AI leaders call for tougher protections against AI-aided bioweaponsRival AI CEOs unite to urge Congress on biosecurity regulations.
- Unlikely alliance: Anthropic's Dario Amodei, OpenAI's Sam Altman, and Microsoft's Mustafa Suleyman signed a joint letter to lawmakers.
- Specific threat: They warn AI could help develop biological weapons that might trigger a global pandemic.
- Regulatory gap: The leaders want Congress to close what they call an 'alarming biosecurity gap' in current AI oversight.
AI Has Come for Serif FontsAI companies increasingly use serif fonts to appear more human.
- Design trend: AI companies are abandoning clean sans-serif for serif fonts to project humanity and trustworthiness.
- Critical pushback: Design critics are calling this aesthetic shift 'tasteslop'—a calculated attempt to manipulate perception.
- Branding strategy: The shift represents a broader move by AI companies to appear less sterile and more approachable.