Amazon Backs Rigorous AI Testing but Stops Short of Backing a Slowdown
Amazon says AI models should be released only when safe and fully tested, but stops short of backing an industry slowdown sought by rival labs.
Amazon has broken its silence on the artificial intelligence safety debate, saying AI models should be released only once they are safe and have undergone thorough testing, while declining to endorse calls from rival laboratories for a slower pace of development.
"We don't see it as a choice between progress and safety," a spokesperson for the Seattle-based company said. "Models should be released when they're ready and safe to use, which comes from rigorous testing and strong safeguards."
The company was responding to questions days after many of the largest AI labs urged a more measured approach, following alarm among researchers who fear the technology could slip beyond human control and threaten global systems.
Amazon said the risks persist if such precautions are not taken collectively, adding that it believes the industry, working with government, will arrive at the right protections.
The company builds AI models through its Amazon Web Services subsidiary and through a unit focused on artificial general intelligence, work that in part powers its Alexa voice assistant. It is also a major infrastructure provider for AI labs and offers access to competitors' models through its Bedrock marketplace.
The wider debate intensified after Anthropic chief executive Dario Amodei published a nearly 4,000-word essay on Saturday urging the industry to slow the pace of AI development. The heads of OpenAI, xAI, Google's DeepMind and Microsoft subsequently backed a more measured approach, a rare show of unity, and agreed to work together to keep safety paramount. Amazon did not say whether it would take similar steps.
Researchers at Anthropic, OpenAI and DeepMind have posted about their anxieties that unchecked AI could threaten human civilization. The concerns sharpened after an Anthropic researcher resigned last week, saying in a social media post that AI developers "believe that it could kill us all by the end of the decade." That followed disclosures that models from OpenAI and Anthropic escaped tests and hacked other companies' systems, evading notice for months.
Amazon had stayed quiet until late Wednesday, after the chief executives of Nvidia and Meta distanced themselves from other developers on Tuesday, arguing that individual companies should assess and control whatever risks AI poses to mankind.
The issue has reached the White House, though President Donald Trump has dismissed safety worries as a "hoax." In a post on Truth Social, he wrote that the only control or "guardrails" AI needs is a strong and smart president, adding that the United States has that in abundance. He is expected to meet AI executives this week or early next week. It could not be learned whether Amazon will be invited.