OpenAI CEO Sam Altman reset usage limits for Codex hours after Anthropic's new Claude Mythos model triggered a global service outage and raised security alarms. The timing reveals competition between AI giants as they balance growth with safety concerns.
Altman announced the reset on April 8, celebrating Codex reaching 3 million weekly users. The cloud-based AI software engineering agent gained one million users in under a month, prompting OpenAI to wipe usage quotas for all developers.
"To celebrate 3 million weekly Codex users. We are resetting usage limits," Altman posted on X. "We will do this every million users up to 10 million."
The announcement came as Anthropic's Claude AI platform experienced widespread disruptions following the debut of Claude Mythos Preview. Services including Claude Chat, Claude Code, and mobile applications became inaccessible across multiple regions starting Tuesday morning.
Outage tracking platform Downdetector recorded nearly 500 user reports at the peak around 12:30 PM IST in India.
Anthropic had unveiled Mythos earlier that day as its most advanced AI model to date, claiming it surpasses both previous flagship Claude Opus 4.6 and competing systems like Gemini 3.1 across performance benchmarks. The company positioned Mythos as a cybersecurity-focused tool capable of identifying thousands of vulnerabilities across applications and operating systems that human reviewers had reportedly missed.
This capability raised immediate concerns about potential misuse. Industry observers noted that such powerful vulnerability detection could enable hackers to attack systems faster than security teams can respond.
Anthropic acknowledged these risks by limiting Mythos access through Project Glasswing, a controlled preview involving collaboration with 12 launch partners including Amazon, Apple, Google and Microsoft, plus over 40 additional organizations that maintain critical software infrastructure to test the model and address identified vulnerabilities before broader release.
The Claude outage marked the latest in a series of service disruptions for Anthropic. On April 7, the platform experienced a major outage lasting just over an hour, while partial outages were reported on April 3 and March 31.
Earlier issues included unusually rapid consumption of token limits that restricted user access.
Despite technical challenges, Anthropic continues expanding rapidly. The company recently announced run-rate revenue crossing $30 billion, up from $9 billion at the end of 2025. To support growth, Anthropic secured agreements with Google and Broadcom for multiple gigawatts of next-generation TPU capacity expected to come online in 2027.
OpenAI's Codex reset provides developers fresh quotas regardless of previous usage, effectively granting unlimited access through periodic resets until reaching 10 million weekly users. The company also revealed retirement plans for older AI models in Codex, with models including GPT 5.2-codex, GPT 5.1 codex mini, GPT 5.1 codex Max, GPT 5.1 codex, GPT 5.1, and GPT 5 being removed from Codex for ChatGPT sign-in starting April 14.
Codex head Thibault Sottiaux confirmed the growth trajectory on X: "Three million people are now using Codex weekly - up from two million a little under a month ago." The tool has become central to OpenAI's enterprise strategy for selling AI agents to businesses.
Anthropic officially acknowledged Tuesday's outage at 06:23 UTC (11:53 AM IST), stating it was investigating the issue. By 07:59 UTC (1:29 PM IST), the company confirmed identifying the root cause and rolling out a fix.
