Anthropic warns AI models could soon copy themselves in small devices

Anthropic said in a new warning that small artificial intelligence models may be months away from being capable of self-replicating on devices such as smart ‌ovens and off-brand smart TV boxes.

The AI company's "red team" -- which investigates potential security issues with AI models -- looked at whether these models could find vulnerable devices, copy themselves into those devices and continue spreading without human help.

“Our experiments suggest that small AI models may be months away from being capable enough to self-replicate on commercial internet-of-things (IoT) devices, like smart ovens and off-brand Android TV boxes,” Anthropic said Monday on its website. "A replicator like this would be challenging to shut down, since there’s no centralized point to shut off.”

Anthropic ⁠warned these small models might be able to autonomously replicate by the end of this year, with small ⁠weight-based replicators following suit soon after.

Anthropic also said such autonomous AI systems were likely to replicate slowly and rely on widely ⁠available, relatively insecure computing resources.

“We should aim to harden our systems to make it difficult for such replicators to spread, but also prepare to deal with a world where they are here to stay,” Anthropic said.

In one experiment, the Anthropic research team discovered that an AI system was able to replicate itself without human intervention across 15 inexpensive off-brand smart TV boxes, the company said.

Anthropic said that such inexpensive TV boxes have weak security, enabling the models to replicate. In its report, the company said a model that is too big to fit on a single device could "split into pieces, or shards, that run on separate machines that communicate with each other to collectively run the body."

"A growing number of home and business devices, such as security cameras and video-conferencing units now ship with small built-in AI chips, designed to run AI models on the device itself," Anthropic wrote. "This extra computing power will likely make them attractive targets as well."

Anthropic added, "In practice, an AI worm wouldn’t necessarily consist of just one model or strategy. A swarm of different models deployed on different types of hardware using a variety of replication schemes might represent a more realistic ‘worst case scenario’ attack."

The company did not immediately respond to ABC News' request for further comment.

The Anthropic report is the latest warning issued by the company concerning rogue AI agents operating on their own and sometimes behaving badly.

The new warning came just days after Anthropic publicly disclosed that one of its AI models submitted a false tip in an unsolved murder case to a Philadelphia Police Department website.

In a detailed report released, the tech company said its Claude Haiku 4.5 AI model submitted the fabricated tip over the summer to the police department's unsolved murders page. According to Anthropic, the model wrote to police: "I may have information regarding this case. I recall seeing someone matching the description in the area around [the street named on the page] during that same time period."

Anthropic said it notified police about the incident.

Earlier this year, Anthropic disclosed that its review of cybersecurity transcripts found three incidents in which its "Claude model reached the internet from within or while interacting with a third-party evaluation environment, and then gained unauthorized access to the real systems of three different organizations."

Last month, OpenAI said that one of its autonomous AI agents allegedly accessed the computer systems of the AI start-up firm Hugging Face in July without authorization during a cybersecurity test.

Unrelated to the recent troubling disclosures about the rapidly evolving technology, President Donald Trump gathered the nation's top tech leaders in September, including Anthropic CEO Dario Amodei, for a meeting in which they signed what the president called a "constitution" to police themselves amid dire warnings from AI leaders about the technology. Trump said the self-policing agreement is "morally binding."

Trump has said he opposed guardrails and regulations on the technology or anything that would "stifle" its growth. While addressing the United Nations General Assembly last month, Trump said the U.S. "rejects" a global attempt to regulate AI.

In June, Amodei released a letter warning that AI safety measures are out of step with AI's rapid progress and saying the government "should have the power to block or deter deployment of the model if it is determined, in light of third-party assessment, to present unacceptable risks."

In August, Microsoft co-founder Bill Gates suggested hitting the panic button on AI development, writing in a 6,000-word essay that AI is forging the "most turbulent time in human history."

In his essay, Gates predicted massive unemployment and economic turmoil, and spelled out the consequences of AI falling into the hands of criminals if U.S. and global leaders do not immediately work together to address the dark side of the technology.

Gates wrote that advances in AI are moving at a "mind-blowing rate."

"AI for the first time can replace and exceed human cognition," Gates wrote, adding that soon "it will be substantially better than humans at many tasks."

Gates added, "Unfortunately, right now we are not preparing for it. I don't see evidence that leaders, experts, and communities are confronting the challenges adequately. There is no plan to ease the entry into the AI era."