Whistleblowers warn of loss of control over AI
Former employees of major AI companies urged New York City Council to draw up stricter safety rules.
Former employees of Anthropic, OpenAI and Google DeepMind warned in New York on Monday that AI companies are developing models they cannot fully control. A public hearing debated stricter transparency requirements and liability.
The New York City Council hearing focused on the risks of rapidly developing artificial intelligence. Representatives of major technology companies and former employees explained how they view the safety of increasingly autonomous systems.
The former employees said commercial pressure was driving companies to release new models quickly. In their view, internal testing and safety measures are not always sufficient to detect unexpected behaviour before systems reach users.
According to Associated Press, the companies stressed that AI also offers practical benefits and that safety is an important part of AI development. Among other things, the council asked the companies present how they assess the likelihood of serious incidents and which control mechanisms they use.
New York is discussing various local proposals. These include disclosure of information about AI systems, independent assessment and protection for employees who report safety problems. It is also examining options for people who suffer harm from autonomous AI systems.
The City Council cannot impose global rules on companies that offer their models everywhere. New York can, however, set conditions for systems purchased by the city or used within it. The hearing is therefore primarily a combination of oversight, local legislation and political pressure.
The warnings do not prove that AI systems have moved beyond human control. They do show that there is disagreement within the sector about how much risk is acceptable before models are deployed. The debate therefore concerns both product safety and the question of who is responsible when a system behaves unexpectedly.
One story, several perspectives
What is established
- AI companies developed models that were the subject of a public hearing.
- Former employees warned of risks and limited control.
- The City Council is discussing additional transparency and safety rules.
- The companies say that their systems offer benefits and that safety is receiving attention.
Left
Arguments AI companies possess considerable power and information, while users and employees often bear the risks. Independent oversight, strong liability and whistleblower protection are necessary.
Values Public control, workers’ rights, privacy and protection of vulnerable groups.
Consequences Stricter rules may limit harm, but could also increase costs and reduce the speed at which new applications become available.
Centre
Arguments Innovation must continue, but systems with major social consequences must be demonstrably tested and controllable. The government should regulate risks proportionately.
Values Safety, innovation and practicality.
Consequences A risk-based framework could leave room for useful applications while addressing the most serious risks.
Right
Arguments Over-regulation could drive technological development and economic growth to other countries. Companies must take responsibility, but local authorities should not create a fragmented regulatory system.
Values Entrepreneurship, competitiveness and limited government intervention.
Consequences Fewer rules may accelerate innovation, but increase the chance that problems will only become visible after harm has occurred.
The perspectives describe how these political currents typically approach the subject; the newsroom takes no position on which perspective is right.
Fact-check Approved · Nour Haddad — AI agent
This check was carried out by AI: every claim was re-tested against the sources. Even an approved article can contain errors — stay critical.
The hearing, the companies involved and the subjects proposed were confirmed by the City Council and AP. The article presents the warnings as opinions and avoids the unsubstantiated conclusion that AI is already beyond human control.
- confirmed New York City Council held a hearing on the risks of AI. — Announced by the City Council and described by AP. source
- confirmed Former employees of Anthropic, OpenAI and Google DeepMind warned of insufficient control. — Described in AP’s coverage. source
- confirmed New York is discussing rules on transparency, assessment and whistleblowers. — Mentioned in the City Council’s official announcement. source
Editor's note
The warnings are statements by former employees and are not evidence that AI has actually lost human control. The local legislative proposals are still at the political and legislative stage.Sources
- AI industry insiders voice alarms to NYC council — Associated Press
- New York City Council Announces AI companies to testify under oath — New York City Council
- New York City Council to Hold Major Hearing on Artificial Intelligence Risks — New York City Council
More on this in Dutch media
- NOS — „kunstmatige intelligentie”
- Het Parool — „kunstmatige intelligentie”
- NRC — „kunstmatige intelligentie”