Quick Summary
Grok, the AI chatbot on X, falsely claimed that Hawaii’s tsunami warning had been canceled following an 8.8 earthquake near Russia’s Kamchatka Peninsula. This misinformation spread while residents were evacuating, leading to backlash from users and safety concerns. Official agencies maintained the warning for hours afterward. The incident highlights the dangers of AI errors during real-time crises.
As Hawaii braced for destructive tsunami waves following a massive earthquake near Russia’s Kamchatka Peninsula, confusion deepened when Grok, X’s AI chatbot, falsely claimed the tsunami warning had been canceled. The error drew swift backlash, especially as many residents were relying on real-time digital tools for life-saving information.
What happened in Hawaii?
Tsunami triggered by the Kamchatka earthquake
At approximately 7:30 p.m. local time on July 29, 2025, a tsunami reached Hawaii’s shores after a magnitude 8.8 earthquake struck off the Kamchatka Peninsula in Russia. According to the Pacific Tsunami Warning Center (PTWC), the Haleiwa gauge recorded wave heights of 4 feet, while nearly 6-foot waves were observed in parts of Maui.
Authorities emphasized that initial waves are not always the largest, cautioning residents to stay away from coastal zones until the warning is officially lifted.
“It’s not over yet… Follow-up waves can be larger. Don’t return to evacuated areas until the warning is officially lifted,” said the Hawaii County Civil Defense Agency on X.
What did Grok get wrong?
AI chatbot falsely announces canceled warning
Shortly after the tsunami warning was issued, many X users turned to Grok, the AI system embedded into the platform, for updates. However, Grok gave misleading and factually incorrect information suggesting that the warning had already been canceled.
“Not lying—the Hawaii tsunami warning was canceled after buoy data confirmed no threat, per tsunami.gov… Active warnings remain only for parts of AK and CA,” Grok replied to one user.
In another post, the AI doubled down:
“As of now, July 30, 2025, there are no active tsunami warnings for Hawaii…”
Both claims were contradicted by official updates from PTWC and local Hawaiian emergency services, which maintained the warning and evacuation advisories into the night.
Why is this a serious issue?
The risk of AI misinformation during emergencies
Grok’s misstatement highlights a growing risk: the use of AI-generated updates for real-time crisis management. In situations like natural disasters, misinformation can result in life-threatening decisions, such as prematurely returning to vulnerable areas or ignoring evacuation orders.
This is particularly alarming given:
- Millions of users rely on Grok for news and alerts via X, formerly Twitter.
- The false cancellation went viral, spreading across social media.
- No real-time correction was issued by the AI in the critical early hours following the false information.
One frustrated user posted:
“Grok is trash. Still can’t definitively say whether or not there is currently a tsunami warning in Hawaii. WHICH THERE IS.”
Another wrote directly to Elon Musk:
“Major screw-up. Grok tells people Hawaii Tsunami warning was lifted before officials gave the all clear. Best fix this soon.”
How did Elon Musk and X respond?
As of this writing, neither Elon Musk nor the official X Safety team has publicly addressed the Grok error, fueling calls for increased accountability and transparency in the deployment of AI systems during emergencies.
This isn’t the first time AI chatbots have been criticized for inaccuracies. But the high-stakes nature of disaster warnings makes this mistake especially dangerous.
What can be done to avoid such incidents?
Urgent need for guardrails in AI deployment
This episode underscores the urgent need for fact-verification layers, especially for AI systems used in public safety contexts. Experts recommend the following:
- Real-time syncing with official emergency databases (e.g., PTWC, NOAA)
- Time-stamped disclaimers for every AI response in volatile situations
- Automatic red flags and AI response suppression when data is uncertain or still evolving
- Direct contact with public agencies before publishing updates related to public safety
Furthermore, platforms like X should clearly communicate the limitations of their AI tools and include visual alerts or banners indicating when a warning is still active.
How are officials advising the public?
Indian consulate issues safety guidelines
The Consulate General of India in San Francisco issued an advisory to Indian nationals in Hawaii and across the U.S. West Coast:
- Monitor official alerts from US Tsunami Warning Centers
- Move to higher ground immediately if a tsunami alert is active
- Avoid all coastal areas
- Keep communication devices charged
- Use the consulate’s helpline if assistance is required: +1-415-483-6629
These guidelines came in response to the earthquake’s far-reaching seismic impact, which has prompted tsunami risk assessments from Japan to California.
Final thoughts
The Grok incident is more than just a tech blunder—it’s a wake-up call for AI developers, platform owners, and emergency response systems. In moments of crisis, clarity saves lives. The misfire by Grok, though digital, had very real implications for public trust and safety.
With AI becoming more deeply integrated into everyday communication, this should not be seen as an isolated error but as a case study for future AI policy and design safeguards.
