‘Careless use of AI is the real threat – not the ghost stories’
Meredith Whittaker, chief executive of the Signal Foundation, challenges the prevailing narrative that artificial intelligence poses an existential threat to humanity, arguing instead that the real danger lies in careless deployment of existing technology. She cites a US intelligence chatbot that fabricated information about a Chinese vessel carrying nuclear material, nearly triggering a military conflict with China, as evidence that poorly implemented AI systems pose more immediate risks than theoretical AI rebellion scenarios.
Named by Time magazine as one of the most influential figures in AI, Whittaker has a track record of opposing military applications of artificial intelligence, including Google's Project Maven for drone targeting, and she has drawn a direct line from that programme to a US missile strike on an Iranian school this year. Signal, which she leads with an estimated 40-70 million users, emphasises encryption protections, putting Whittaker in direct conflict with governments pushing for weaker encryption in the name of protecting children.
- Real AI danger is careless use, not systems turning against humans.
- US chatbot almost triggered war with China by inventing false intelligence.
- Signal CEO opposes government pressure to weaken encryption for surveillance.
New here? Start with this
Meredith Whittaker is chief executive of Signal, a messaging application used by tens of millions of people to keep conversations private and secure. She has been named one of the most influential figures in artificial intelligence and has long opposed military uses of AI technology, specifically Google's Project Maven for drone targeting and connecting such programmes to a US missile strike on an Iranian school earlier this year.
The story centres on Whittaker's challenge to a widespread view that artificial intelligence could become an existential threat—so powerful it could endanger humanity. Instead, she argues the real danger comes from careless deployment of AI systems already in use today. She cites a US intelligence chatbot that fabricated information about a Chinese vessel carrying nuclear material; the false data nearly triggered military conflict with China.
This debate matters because it shapes how governments regulate artificial intelligence and allocate resources to AI-related policy. Whittaker's position also intersects with ongoing tensions between governments seeking to weaken encryption in messaging apps, which they argue is necessary for public safety, and privacy advocates who insist strong encryption is essential for protecting users.
Both sides, in good faith
The strongest fair case each way — we don't pick a winner.
The case for
Whittaker's position emphasises that artificial intelligence is already causing documented harm through hallucinations, biased outputs, and poor military deployment—risks that demand urgent governance now rather than speculation about hypothetical superintelligence. She argues that finite resources for AI safety should address the immediate geopolitical dangers from careless systems, such as intelligence chatbots triggering near-conflicts, rather than diverting attention to theoretical existential scenarios that may never materialise. This pragmatic focus on real, preventable harms reflects responsible stewardship of technology already in widespread use.
The case against
Existential risk advocates contend that transformative AI could eventually pose civilisation-scale dangers that justify proactive research and governance, just as nuclear and pandemic risks warrant prevention before catastrophe. They argue that waiting for catastrophic incidents to spur action may be too late if the risks are truly existential, and that building safety practices and institutional competence now is essential preparation for more powerful systems. Near-term and long-term AI safety work need not compete; investing in both addresses the full spectrum of risks, and the magnitude of potential future harms—however uncertain in probability—warrants sustained attention alongside immediate concerns.