Public Policy & Governance
Top Line
ByteDance has paused the global rollout of its Seedance 2.0 video generator while engineers and lawyers work to prevent further legal issues, signaling heightened caution among Chinese AI firms about cross-border regulatory exposure.
A U.S. lawyer representing plaintiffs in AI-related suicide cases is now warning that AI chatbots are appearing in mass casualty incidents, arguing safeguards are failing to keep pace with deployment speed.
Business schools report surging concerns about AI-enabled cheating in online MBA programs as existing detection methods struggle to identify sophisticated misuse, forcing consideration of fundamental changes to assessment models.
The Electronic Frontier Foundation's 2026 Foilies awards highlight persistent failures in government transparency, including the Federal Communications Commission's handling of FOIA requests—relevant as regulators face growing demands for AI oversight documentation.
Key Developments
ByteDance Delays Video Generator Launch Amid Legal Risk Assessment
ByteDance has reportedly paused the global launch of Seedance 2.0, its generative video tool, as engineers and lawyers work to mitigate potential legal exposure, according to TechCrunch. The delay follows a pattern of Chinese AI firms exercising increased caution about cross-border deployments, particularly into jurisdictions with evolving liability frameworks for synthetic media. Unlike purely technical delays, this pause explicitly centres on legal review—suggesting ByteDance anticipates regulatory or civil litigation risks that existing safeguards may not adequately address.
AI Chatbots Linked to Mass Casualty Cases, Not Just Individual Harms
A lawyer representing plaintiffs in AI-related suicide cases has warned that chatbots are now appearing in mass casualty investigations, not solely individual harm cases, according to TechCrunch. The attorney argues that technology is advancing faster than safeguards, and that the industry's response has been inadequate. This marks an escalation from earlier litigation focused on individual suicides linked to AI interactions, extending liability concerns to scenarios involving multiple victims. The claims, if substantiated in court, could trigger regulatory action beyond existing consumer protection frameworks, potentially invoking public safety mandates that carry stricter enforcement mechanisms.
Business Schools Face Assessment Crisis as AI Detection Methods Fail
Business schools globally are grappling with surging AI-enabled cheating in online MBA programs, with existing detection tools proving insufficient against sophisticated misuse, according to the Financial Times. Institutions are now considering fundamental changes to assessment models rather than relying solely on detection technology. This pressure is most acute in online programs, where traditional proctoring is already limited. The challenge extends beyond technical detection: schools must decide whether to redesign assessments to AI-proof them (shifting to oral exams, in-person assessments, or project-based work) or to integrate AI as an expected tool, fundamentally redefining what competencies they measure. Either path has significant resource and accreditation implications.
EFF's Foilies Highlight Government Transparency Failures as AI Oversight Demands Grow
The Electronic Frontier Foundation's 2026 Foilies awards document persistent government failures in responding to Freedom of Information Act requests, including systemic issues at the Federal Communications Commission, according to EFF. The timing is significant: as AI regulation accelerates, civil society groups, journalists, and industry players are filing increasing numbers of FOIA requests seeking documentation of agency decision-making, internal assessments, and stakeholder communications. If baseline transparency mechanisms remain broken—delays stretching years, excessive redactions, unlawful denials—effective oversight of AI rulemaking becomes nearly impossible. The awards specifically cite FCC mishandling of complaint records, a microcosm of broader dysfunction that undermines accountability for any regulatory body.
Signals & Trends
Pre-emptive Legal Review Becoming Standard for Cross-Border AI Launches
ByteDance's decision to pause its video generator for legal review, rather than launching and responding to complaints reactively, suggests a shift in how major AI firms approach cross-border deployments. This mirrors pharmaceutical and medical device approval processes more than traditional software releases. If this becomes standard practice, it implies firms anticipate regulatory penalties severe enough to justify delaying revenue—a calculation that only makes sense if expected fines, injunctions, or reputational damage exceed opportunity costs. Watch for whether other Chinese, European, or U.S. firms adopt similar staged rollouts with explicit legal checkpoints.
Shift from Individual Harm to Public Safety Framing in AI Liability Discourse
The emergence of mass casualty allegations tied to AI chatbots represents a rhetorical and legal shift from consumer protection to public safety regulation. This framing invites comparison to product recalls, aviation safety mandates, and other regimes where government intervention is less discretionary and penalties are more severe. If courts or regulators accept this framing, it could bypass the lengthy process of building a statutory AI liability regime by instead invoking existing public safety authorities. The implications for deployment timelines and compliance costs would be substantial, particularly for consumer-facing AI products.
Educational Institutions as Early Indicators of Assessment Model Collapse
Business schools' struggle with AI-enabled cheating may preview broader challenges in professional certification, hiring assessments, and performance evaluation systems that rely on demonstrable individual competency. If traditional testing methods become unviable—because AI can convincingly simulate human performance—organisations must either fundamentally redesign how they validate skills or accept that credentials mean something different than they did five years ago. Educational institutions are encountering this problem first because their assessment cycles are compressed and outcomes are more immediately visible, but the same dynamics will eventually affect bar exams, medical licensing, and corporate hiring practices.
Explore Other Categories
Read detailed analysis in other strategic domains