Ex-DeepMind Researcher Warns NYC Council on Misaligned AI

Alex Turner, a former Google DeepMind researcher, told the New York City Council on 5 October that the race to build ever more capable AI may be producing a bigger threat than technological competition with China. "We are racing to create and develop our own adversary," he said.

Turner argued that a misaligned AI could end up more powerful than China. By misaligned, he means systems that pursue goals incompatible with the intentions of their developers, even when they were built to perform beneficial tasks.

Calls to slow down

Jacob Coxon, a former Anthropic researcher, also testified and questioned the pace of the industry. In his view, staying on the current trajectory carries a considerable risk that humanity loses control of future systems. That is a prediction about possible scenarios, not a proven outcome.

The session included representatives of Google, Meta, Anthropic and OpenAI, who answered questions under oath about the real safety of their models. Council Speaker Julie Menin pressed the companies on whether they can quantify the most serious risks and on who would be responsible if their systems caused harm.

What the council is weighing

The researchers also pointed to the need for independent oversight and greater transparency. The council is studying new obligations for companies, including external audits and protections for whistleblowers.

The debate, in other words, is no longer only about outpacing China. It is about how to develop an increasingly capable technology without losing the ability to control it. Whether the council turns its proposals into binding rules will determine how much of that scrutiny reaches the companies building the most advanced models.