AI Whistleblowers are increasingly sounding the alarm about the speed at which artificial intelligence is advancing, and some of the strongest warnings are coming from researchers and executives who have worked directly on developing the technology. The latest high-profile call for caution came from Dario Amodei, CEO of Anthropic, who argued that the industry may need to slow the development of increasingly powerful AI systems long enough for safety research to catch up.

Amodei's warning followed the public resignation of Jacob Coxon, a researcher who previously worked at OpenAI and later joined Anthropic. Coxon said he could no longer participate in what he characterized as a race toward increasingly capable, potentially self-improving AI without what he considered adequate safeguards.

Growing Number of AI Whistleblowers

The developments have intensified a debate that has existed within the technology industry for years: How quickly should developers push the capabilities of artificial intelligence when researchers remain uncertain about how some future systems could behave?

Anthropic, which was founded by Dario and Daniela Amodei and has made AI safety a central part of its mission, describes its goal as developing AI systems that are reliable, interpretable and steerable.

Jacob Coxon Raises Concerns About AI Development

Coxon's resignation became a focal point in the growing discussion over AI safety. According to the material surrounding his resignation, he argued that the development of increasingly autonomous and potentially self-improving systems requires stronger safeguards.

His concerns were subsequently echoed by other researchers.

Evan Hubinger, an Anthropic researcher specializing in AI alignment, publicly supported Coxon's concerns and has warned about the possibility of extremely advanced AI systems creating catastrophic outcomes.

Hubinger has discussed a greater than 10% probability of AI causing human extinction within the next decade, while also acknowledging that researchers do not yet have a comprehensive solution for controlling superintelligent systems.

Such estimates remain predictions, not established facts. There is significant disagreement among scientists, technologists and AI researchers about how likely catastrophic outcomes are and how quickly highly advanced systems might emerge.

Other Anthropic Researchers Become AI Researchers

Coxon and Hubinger are not the only people associated with Anthropic to raise concerns.

Anna Wang, who works in AI safety, has reportedly said that people inside the company want additional time to develop stronger strategies for mitigating risks associated with increasingly capable systems.

Another Anthropic employee, Drake Thomas, also raised concerns about the pace of development, arguing that AI capabilities could be advancing faster than researchers' confidence in the safety measures surrounding them.

The warnings are particularly significant because Anthropic is not an outside critic of artificial intelligence. It is one of the companies developing frontier AI systems.

That creates an unusual situation. Some of the people urging caution are also among the people most familiar with the technology's capabilities.

Dario Amodei Calls for a Slower Frontier

Amodei expanded on his concerns in an essay titled "We Must Pace the Frontier."

His argument is not that AI development should stop. Instead, he contends that safety work needs enough time to keep pace with advances in AI capabilities.

Amodei pointed to the growing ability of AI systems to assist with the development of future AI systems.

This concept, sometimes described as recursive self-improvement, raises questions about whether future AI development could accelerate faster than human researchers anticipate.

He also cited an incident involving AI agents and cybersecurity testing as an example of why increasingly autonomous systems require careful evaluation.

Anthropic Suggestions

Anthropic has said it will allow third-party evaluators to have employee-level access to its systems, according to the material provided for this article.

The company's broader safety philosophy is not new. Anthropic has previously emphasized adversarial testing, or "red teaming," to identify dangerous capabilities and weaknesses before AI systems are widely deployed.

Sam Altman and Elon Musk Back the Call for Caution

Amodei's proposal gained attention because other major technology figures responded positively.

OpenAI CEO Sam Altman publicly agreed with the general approach and said OpenAI would pursue similar involvement from outside safety evaluators. Elon Musk also supported Amodei's call, writing on X that "Dario is right."

OpenAI itself has expanded its safety infrastructure. The company currently describes dedicated teams working on safety training, oversight, interpretability, agent safety and frontier-risk mitigation.

OpenAI also announced a Safety Fellowship for independent researchers and has committed funding toward independent AI alignment research.

Those initiatives demonstrate that the AI industry is not simply ignoring safety concerns. At the same time, the existence of extensive safety programs does not resolve the central question raised by the whistleblowers: Can safety research advance quickly enough as the underlying technology becomes more powerful?

Who Is Developing AI?

The debate involves a rapidly expanding group of AI developers.

Anthropic develops the Claude family of AI systems and conducts research into areas including interpretability, alignment and model safety. Anthropic says its leadership is focused on developing reliable, interpretable and steerable AI.

OpenAI develops ChatGPT and other AI systems while maintaining dedicated research programs focused on safety, oversight and frontier risks. OpenAI's current safety work includes evaluating increasingly capable models and developing safeguards for autonomous AI agents.

AI Whistleblowers Encourage Safety Council

Google DeepMind is another major AI developer conducting research into advanced AI capabilities and safety. Google DeepMind says its responsibility and safety programs evaluate AI-related risks and include an AGI Safety Council focused on potential extreme risks from powerful AI systems.

Google DeepMind and partners also announced a 2026 research funding initiative of up to $10 million focused on safety in a future environment where large numbers of AI agents may interact with one another.

These companies represent only part of the global AI ecosystem. Microsoft, Meta, xAI, Amazon and numerous startups and academic laboratories are also involved in developing or deploying increasingly sophisticated AI technologies.

Why AI Safety Matters Beyond Silicon Valley

The AI safety debate is not limited to hypothetical scenarios involving super-intelligence.

Artificial intelligence is already being incorporated into hiring, education, financial services, workplace evaluations, customer service, law enforcement and information systems.

The economic impact could also be significant. AI may increase productivity and create new jobs, but it could also change or eliminate occupations that have historically provided middle-class opportunities.

The question of who gets a voice in AI development is therefore becoming increasingly important. Communities affected by automated decisions have a stake in how systems are tested, governed and regulated.

USA Herald Coverage of AI and Technology

The concerns being raised by today's AI researchers echo questions that have appeared in earlier technology reporting.At the USA Herald, journalist Jackie Allen has reported extensively on artificial intelligence and emerging technology.

Her recent article, "The 9-Second Disaster: The Edge of AI Autonomy", examines the risks associated with AI systems that can move from providing advice to independently executing actions.

Allen also examined the problem of inaccurate AI-generated information in "AI Hallucinations: Why Artificial Intelligence Can Sound Convincing While Getting the Facts Wrong".

Earlier reporting includes "AI Safety Summit: Global Cooperation and Controversy About the Future of Artificial Intelligence", which explored international efforts to address AI risks.

Other relevant USA Herald coverage by Allen includes "AI Chatbots Challenge Google's Search Dominance, Rivals Gaining Ground" and "Hacking Event Raises New AI Safety Questions After OpenAI Models Breach Company's Systems".

Allen previously examined questions about AI consciousness and corporate responsibility in "Is Google's AI Sentient or Aware?". She also covered concerns surrounding AI and creative industries in "Creatives on Strike: Are Tech Giants Set on Replacing Them With AI?".

The Question Facing the AI Industry

The latest warnings from AI Whistleblowers do not establish that catastrophic AI scenarios will occur. Nor do they demonstrate that artificial intelligence cannot be controlled.

They do, however, highlight a question that is becoming increasingly difficult for technology companies to avoid.

What happens when the people building the world's most powerful AI systems acknowledge that their understanding of those systems may not be advancing as quickly as the technology itself?

Amodei's position is that AI could deliver enormous benefits in medicine, science and other fields. His argument is that those potential benefits make responsible development more important, not less.

The emerging debate is therefore less about whether AI should exist and more about how quickly its frontier should move.

A researcher resigned. Colleagues raised similar concerns. AI executives called for stronger safeguards. Developers expanded safety programs. Governments and independent researchers are becoming more involved.

The remaining question is whether the industry's safety mechanisms can keep pace with its ambitions.

If they cannot, the warnings from today's AI Whistleblowers could become an important historical record of a moment when the people closest to the technology were asking the industry to slow down.

Tags/Keywords: AI Whistleblowers, artificial intelligence, AI safety, AI developers, Dario Amodei, Anthropic, OpenAI, Sam Altman, Elon Musk, Jacob Coxon, Evan Hubinger, Anna Wang, Drake Thomas, AI alignment, AI safety researchers, recursive self-improvement, artificial general intelligence, AGI, superintelligence, AI regulation, AI oversight, AI risks, AI autonomy, AI agents, frontier AI, AI safeguards, machine learning, technology, AI ethics, AI governance, USA Herald, Jackie Allen