The company is recruiting 32 specialists to test whether advanced artificial intelligence models could be misused for explosives, nuclear weapons, cybercrime and other potentially catastrophic activities.
By yourNEWS Media Newsroom
Anthropic is seeking nearly three dozen artificial intelligence safety specialists to assess whether its most advanced models could help malicious users develop explosives, nuclear weapons, cyberattacks, financial scams or other dangerous capabilities, according to a Daily Caller News Foundation report.
The AI company has posted 32 safety-related openings as concerns grow throughout the technology industry about models that can analyze complex scientific information, locate software vulnerabilities and generate detailed instructions at speeds that were previously impossible.
The available positions include analysts specializing in radiological and nuclear threats, explosives, cybercrime and financial fraud, according to a report examining Anthropic’s hiring effort.
Some of the jobs pay more than $200,000 annually.
“As an Enforcement Analyst focused on Radiological & Nuclear Harms, you will play a critical role in protecting against the misuse of AI systems for radiological and nuclear harms,” one listing stated.
Anthropic’s safety analysts will be responsible for stress-testing the company’s models before they are made available to the public. The work involves attempting to identify ways criminals, terrorists or hostile governments could bypass safeguards and obtain information that might support harmful activity.
“Ensuring our models don’t provide potentially harmful information is central to responsible development,” an Anthropic spokesperson said.
“That’s why we regularly hire experts in a wide range of sensitive fields — people who understand these harms and how AI can advance them — to stress-test our systems and bolster our defenses before a model ever goes live,” the spokesperson added.
The hiring campaign reflects warnings from Anthropic CEO Dario Amodei, who has repeatedly argued that rapidly advancing AI systems could create catastrophic risks if they make specialized scientific or technical knowledge broadly accessible.
In a January essay, Amodei identified the possibility of AI-assisted biological weapons development as one of the most urgent safety concerns.
“I do not think biological attacks will necessarily be carried out the instant it becomes widely possible to do so — in fact, I would bet against that,” Amodei wrote.
He nevertheless warned of a “serious risk” that such attacks could eventually cause deaths “potentially in the millions or more.”
Anthropic’s approach has drawn criticism from officials and industry figures who argue that warnings about catastrophic AI threats could be used to justify regulations favoring established companies over smaller competitors.
David Sacks, who served as President Donald Trump’s AI and cryptocurrency czar, accused Anthropic in October 2025 of pursuing a “sophisticated regulatory capture strategy based on fear-mongering.”
Sacks’ criticism suggested that large AI companies could use safety concerns to encourage regulatory requirements that they can afford to meet but that would create obstacles for newer or less heavily funded competitors.
Anthropic has continued to argue that the growing capabilities of AI systems require stronger safeguards before models are widely deployed.
The company called in June for slowing global AI development, saying additional time would allow governments and companies to address the technology’s “immense implications.”
Anthropic also acknowledged a potential danger in unilateral restraint, warning that if “a slowdown simply lets the least cautious actors catch up technologically, it could leave everyone less safe.”
The company’s hiring plans come as other major AI developers expand their own safety operations.
OpenAI, the developer of ChatGPT, is recruiting an analyst specializing in biological and chemical risks, with compensation reaching as high as $445,000 annually.
The position reflects growing concern that advanced models could assist users in analyzing dangerous pathogens, chemical compounds or laboratory procedures.
OpenAI has also reorganized parts of its safety and research structure. Johannes Heidecke, formerly the company’s head of Safety Systems, left OpenAI during that reorganization.
Anthropic’s recruiting effort coincides with an expanding state-level lobbying campaign for stronger AI rules.
The company has continued pressing state governments to adopt more stringent safety requirements, according to a Politico report on its regulatory strategy.
“While there are some in the industry that think of state policy as a way to create a ceiling for federal legislation, Anthropic is not just looking to support the same bill across the country in every single state,” Cesar Fernandez, Anthropic’s head of state and local government relations, told the outlet.
The company’s approach appears to favor state policies tailored to different jurisdictions rather than one identical legislative framework enacted throughout the country.
OpenAI has pursued a different state-level strategy.
Chris Lehane, the company’s top lobbyist, has advocated for states to approve similar regulatory frameworks, allowing policymakers to establish common AI standards without waiting for Congress to overcome its partisan divisions.
The competing approaches illustrate a broader disagreement within the AI industry over how governments should regulate increasingly capable systems.
Anthropic has placed particular emphasis on preventing catastrophic misuse and has argued that AI developers must test models against threats involving biological agents, nuclear materials, cyber operations and other high-risk areas.
Its critics contend that the company’s warnings may exaggerate speculative dangers while creating support for regulations that protect the market position of established developers.
The 32 job listings show that Anthropic is preparing for the possibility that future models could provide dangerous assistance even when developers do not intend them to do so.
Experts hired for the positions would be expected to understand both the relevant scientific or criminal activity and the ways an AI system might accelerate it.
The company’s radiological and nuclear safety analysts, for example, would test whether models reveal restricted or dangerous information, assist with acquiring materials or help users overcome technical barriers.
Specialists in explosives would examine whether safeguards prevent models from generating actionable instructions for constructing dangerous devices. Cybersecurity analysts would test whether users could obtain assistance with finding and exploiting software weaknesses, while fraud specialists would examine potential uses involving financial deception and scams.
Anthropic says the goal is to identify those weaknesses internally and strengthen defenses before a model becomes publicly available.
The recruitment campaign also demonstrates how AI safety has developed into a highly specialized and well-funded professional field.
Compensation exceeding $200,000 at Anthropic and reaching $445,000 for an OpenAI biological and chemical risk position reflects the value companies are placing on experts capable of evaluating rare but potentially devastating threats.
Whether the expansion of safety teams will prevent misuse depends on how quickly models advance, whether safeguards can withstand attempts to bypass them and whether competing developers adopt similar protections.
For Anthropic, the combination of internal safety hiring and state-level lobbying represents a broader strategy to make catastrophic-risk prevention a central part of AI development and regulation.
Original article
Comments