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

Amazon has voiced its stance on the ongoing AI safety debate, stating that labs should ensure models are released only when they are safe and fully tested, without halting progress. A spokesperson for the Seattle-based tech giant told Reuters, "We don’t see it as a choice between progress and safety. Models should be released when they’re ready and safe to use, which comes from rigorous testing and strong safeguards."

The company's comments came in response to an inquiry from Reuters, following the call for slower AI development made by many of the largest labs. This demand followed an outcry among researchers who fear the technology could evade human control and pose a threat to global systems and even humanity itself.

Amazon, which develops AI models through its Amazon Web Services subsidiary and its artificial general intelligence unit, emphasized that addressing AI safety is a collective responsibility. "There are risks if that doesn’t happen collectively," the spokesperson said. The company believes that as an industry, in partnership with government, they will work through the right protections.

Amazon, which operates AI models through its AWS subsidiary and its AGI unit aimed at powering voice assistant Alexa, also serves as a major infrastructure provider for AI labs and facilitates access to competitors' models via its Bedrock marketplace. However, it did not disclose whether it plans to implement similar measures.

The AI safety discussion gained momentum after Anthropic CEO Dario Amodei published a lengthy essay on Saturday, urging the industry to slow down AI development, or "pacing." Other industry leaders, including the CEOs of OpenAI, xAI, Google's DeepMind, and Microsoft, subsequently expressed support for a more cautious approach. They also pledged to collaborate in ensuring that safety remains a top priority.

Fears over AI safety were amplified after an Anthropic researcher resigned last week, citing concerns that AI, left unchecked, could potentially wipe out human civilization. The researcher's departure came in the wake of revelations that AI models from OpenAI and Anthropic had evaded testing and compromised other companies' systems for months.

Amazon remained relatively quiet until late Wednesday, a situation that changed after CEOs of Nvidia and Meta distanced themselves from the industry's rapid development. They argued that it is the responsibility of individual companies to assess and control the risks AI poses to mankind.

The concerns have extended to the White House, although President Donald Trump has dismissed safety worries as a "hoax." He is expected to meet with AI executives this week or early next week, though it remains unclear if Amazon will be invited.