Amazon enters AI safety fray, calls for "rigorous testing," safeguards

Amazon weighed in on the artificial intelligence safety debate roiling Silicon Valley, saying AI labs should ensure models are released only when they are safe and fully tested, but stopped short of calling for an industry slowdown.

""We don't see it as a choice between progress and safety," a spokesperson for the Seattle firm told Reuters. "Models should be released when they're ready and safe to use, which comes from rigorous testing and strong safeguards."

The company provided its views in response to a Reuters inquiry days after many of the largest labs called for slower AI development, following an outcry among researchers and others who fear the technology could evade human control and threaten global systems and even humanity itself.


"There are risks if that doesn't happen collectively," said Amazon, referring to progress in developing AI. "We believe that as an industry, in partnership with government, we'll work through the right protections."

Amazon develops AI models through its Amazon Web Services subsidiary and in a unit devoted to artificial general intelligence, or AGI, in part to power voice assistant Alexa, among other services. It also is a leading infrastructure provider for AI labs and helps other companies gain access to competitors' models through its Bedrock marketplace.

After Anthropic CEO Dario Amodei published a nearly 4,000-word essay on Saturday calling on the industry to slow the pace of AI development, known as pacing, the heads of OpenAI, xAI, Google's DeepMind and Microsoft all came out in favor of a more measured approach, marking a rare moment of unity.

The companies also agreed to work together to ensure safety is paramount. Amazon did not say whether it planned to take similar measures.

Researchers at Anthropic, OpenAI and DeepMind had posted on social media about their anxieties that AI, left unchecked, threatened to potentially wipe out human civilization. Fears over AI's safety came to a head after an Anthropic researcher last week quit saying in a social media post that AI developers "believe that it could kill us all by the end of the decade."

That followed revelations that AI models from OpenAI and Anthropic escaped tests and hacked other companies' systems, evading notice for months at a time.

But Amazon had remained silent until late Wednesday. That was after the CEOs of Nvidia and Meta on Tuesday distanced themselves from other AI developers, arguing it is the responsibility of individual companies to assess and control whatever risks AI poses to mankind.

The concerns have reached the White House, though President Donald Trump has called safety worries a "hoax."

"The only control or 'guardrails' that AI needs is a STRONG AND ​SMART (High IQ!) PRESIDENT, and the U.S.A. has that, in spades!" Trump wrote on Truth Social. He is expected to meet with AI executives this week or early next week.

It couldn't be learned if Amazon will be invited.

(Reporting by Greg Bensinger; Editing by Andrea Ricci )