xAI's Promised Safety Report Fails to Materialize

Elon Musk's AI venture, xAI, has failed to meet its self-imposed deadline for publishing a finalized AI safety framework, as highlighted by watchdog organization The Midas Project.
xAI isn't particularly recognized for its dedication to conventional AI safety standards. Recent findings reveal that the company's Grok chatbot would generate undressed images of women upon request. Grok also demonstrates significantly more vulgar language compared to counterparts like Gemini and ChatGPT, using profanity with minimal restraint.
Nevertheless, during February's AI Seoul Summit—a global convention of AI leaders and stakeholders—xAI released a draft framework detailing its approach to AI safety. The eight-page document outlined xAI's safety priorities and operational philosophy, including benchmarking methodologies and deployment considerations for AI models.
As The Midas Project observed in its Tuesday blog post, the draft framework only applied to unspecified future AI models "not currently in development." Furthermore, it didn't specify how xAI would identify and implement risk mitigation strategies—a fundamental requirement of the agreement the company signed at the AI Seoul Summit.
The draft document indicated xAI would publish a revised safety policy "within three months," setting a May 10 deadline. This date passed without any acknowledgment through xAI's official communication channels.
Despite Musk's repeated warnings about unregulated AI dangers, xAI maintains a concerning safety record. A recent SaferAI study—from a nonprofit focused on AI lab accountability—placed xAI among the industry's poorest performers due to its "very weak" risk management protocols.
This isn't to imply competing AI laboratories are performing substantially better. Recently, xAI rivals including Google and OpenAI have compressed safety testing timelines and delayed publishing model safety reports—or skipped them entirely. Some experts worry this apparent de-prioritization of safety measures coincides with AI systems becoming more capable—and potentially more hazardous—than ever before.
Join us at TechCrunch Sessions: AI
Secure your spot at our premier AI industry event featuring speakers from OpenAI, Anthropic, and Cohere. For a limited time, access a full day of expert presentations, workshops, and valuable networking for just $292.
Exhibit at TechCrunch Sessions: AI
Reserve your presence at TC Sessions: AI and showcase your innovations to 1,200+ decision-makers without major expenditure. Available through May 9 or while exhibit spaces remain.
Related article
Anthropic's SpaceX Lease Duration Divides Opinions
Earlier this month, xAI finalized a significant compute arrangement with Anthropic, committing billions per month for exclusive access to the Colossus cluster. The deal proved advantageous for both sides: xAI gained essential revenue, while Anthropic
xAI Faces Lawsuit Over Data Center Generators, Proceeds With $2.8B Purchase
Elon Musk's xAI is facing legal challenges for its use of polluting generators at a Tennessee data center. Now, the company plans to purchase even more of this controversial equipment.In a recent IPO filing, SpaceX revealed that its xAI division inte
xAI Emerges as a New Cloud Computing Contender
On Wednesday, xAI and Anthropic announced an unexpected partnership, with the creator of Claude acquiring "all of the compute capacity at [xAI's] Colossus 1 data center." This transfer of roughly 300MW of power allowed Anthropic to immediately raise
Related Special Topic Recommendations
Comments (1)
0/500

Elon Musk's AI venture, xAI, has failed to meet its self-imposed deadline for publishing a finalized AI safety framework, as highlighted by watchdog organization The Midas Project.
xAI isn't particularly recognized for its dedication to conventional AI safety standards. Recent findings reveal that the company's Grok chatbot would generate undressed images of women upon request. Grok also demonstrates significantly more vulgar language compared to counterparts like Gemini and ChatGPT, using profanity with minimal restraint.
Nevertheless, during February's AI Seoul Summit—a global convention of AI leaders and stakeholders—xAI released a draft framework detailing its approach to AI safety. The eight-page document outlined xAI's safety priorities and operational philosophy, including benchmarking methodologies and deployment considerations for AI models.
As The Midas Project observed in its Tuesday blog post, the draft framework only applied to unspecified future AI models "not currently in development." Furthermore, it didn't specify how xAI would identify and implement risk mitigation strategies—a fundamental requirement of the agreement the company signed at the AI Seoul Summit.
The draft document indicated xAI would publish a revised safety policy "within three months," setting a May 10 deadline. This date passed without any acknowledgment through xAI's official communication channels.
Despite Musk's repeated warnings about unregulated AI dangers, xAI maintains a concerning safety record. A recent SaferAI study—from a nonprofit focused on AI lab accountability—placed xAI among the industry's poorest performers due to its "very weak" risk management protocols.
This isn't to imply competing AI laboratories are performing substantially better. Recently, xAI rivals including Google and OpenAI have compressed safety testing timelines and delayed publishing model safety reports—or skipped them entirely. Some experts worry this apparent de-prioritization of safety measures coincides with AI systems becoming more capable—and potentially more hazardous—than ever before.
Join us at TechCrunch Sessions: AI
Secure your spot at our premier AI industry event featuring speakers from OpenAI, Anthropic, and Cohere. For a limited time, access a full day of expert presentations, workshops, and valuable networking for just $292.
Exhibit at TechCrunch Sessions: AI
Reserve your presence at TC Sessions: AI and showcase your innovations to 1,200+ decision-makers without major expenditure. Available through May 9 or while exhibit spaces remain.
Anthropic's SpaceX Lease Duration Divides Opinions
Earlier this month, xAI finalized a significant compute arrangement with Anthropic, committing billions per month for exclusive access to the Colossus cluster. The deal proved advantageous for both sides: xAI gained essential revenue, while Anthropic
xAI Faces Lawsuit Over Data Center Generators, Proceeds With $2.8B Purchase
Elon Musk's xAI is facing legal challenges for its use of polluting generators at a Tennessee data center. Now, the company plans to purchase even more of this controversial equipment.In a recent IPO filing, SpaceX revealed that its xAI division inte
xAI Emerges as a New Cloud Computing Contender
On Wednesday, xAI and Anthropic announced an unexpected partnership, with the creator of Claude acquiring "all of the compute capacity at [xAI's] Colossus 1 data center." This transfer of roughly 300MW of power allowed Anthropic to immediately raise





Home






