If you are a keen observer of the developments in AI, you know that between March 27th, 2026, and today (April 22, 2026), a lot has happened.

Here’s a quick recap/timeline:

Around March 26th, 2026, a draft post about Mythos was sitting on Anthropic’s server, and a Fortune reporter discovered it. She got in touch with a couple of researchers from the University of Cambridge to confirm what she had discovered. And so, weeks before Anthropic was ready to announce it, the world came to know about the existence of the then-myth-now-reality creature.

Ten days later, on April 7th, 2026, probably hastened into action by the leak, Anthropic announced Claude Mythos Preview – a frontier AI model it deemed too powerful to release publicly. Project Glasswing was born. It brought together tech majors, including Amazon, Microsoft, Nvidia, and Apple, as well as over 40 organizations that build or maintain critical software infrastructure. The idea was to provide them with early access to Mythos. Why? Because in its testing, Mythos had proved to be a hacking prodigy, with the potential to find and exploit vulnerabilities in existing software systems.

In the next few days, we saw the stock market and even the White House react. The White House invited Amodei for a discussion (remember, when quite recently, the Pentagon had designated Anthropic a supply-chain risk?). The Federal Reserve and Treasury (Powell and Bessent) met with the Heads of the major U.S. banks and discussed the Mythos model’s potential cyber risks. (JPMorgan Chase and Bank of America have been actively testing their systems using Mythos.)

Recently, a major cybersecurity report titled “The AI Vulnerability Storm” was released by the SANS Institute, Cloud Security Alliance, and OWASP GenAI, warning that Mythos was compressing the window between vulnerability discovery and exploitation down to hours.

You can download the 29-page evolving pdf here: https://labs.cloudsecurityalliance.org/wp-content/uploads/2026/04/mythosreadyv92.pdf

Between April 15th and 17th, the regulators and policymakers gathered at the IMF spring meeting in Washington. They issued a series of warnings about the cybersecurity challenges Mythos poses to the banking industry and its legacy systems. Recently (on April 21, 2026), Anthropic announced plans to extend Mythos access to European and UK banks – quite possibly within days. German central bank chief Nagel publicly called for all institutions to have access to Mythos, so that the playing field stays level for all the banks around the globe.

My Thoughts:

Right now, the entire world stands at a precipice. How we handle the immense power of AI is going to define the future of humanity. However, instead of uniting as humans, we are dividing as humans. We’ve always been divided. We’ve been divided as countries, religions, cultures, communities, societies, families…we’ve fought each other since the dawn of civilization.

Three years ago, when AI couldn’t fingers, we laughed at it and said it couldn’t compete with us.

Two years ago, when it hallucinated, we laughed again. “Oh, we need humans to make sure it does stuff.” And we also said, “Hey, it can’t prompt itself. Some of us called ourselves smart because we wrote better prompts than our neighbors.

A year ago, we told ourselves, “AI can’t do stuff on its own. A human-in-the-loop is needed.” We told ourselves, “The smart ones will always be needed,” and we whispered to ourselves that we are the smart ones.

Today, with agentic AI we offload our work to AI by giving it the permissions required to carry out complete tasks for us. Today, a person with moderate coding skills can use an AI with Mythos-like capabilities and rain hell over our financial systems, eventually, impacting millions of people.

And yet, we humans are busy warring, fighting, nagging, and then eating and drinking to handle our depression. We provide human-generated content to the AI through our chats and call it AI usage. We fit cameras over the heads of humans to train the robots.

We can’t be bothered to regulate at a speed that matches the speed of AI development. We can’t be bothered to have an objective data-based conversation on AI and how it’s impacting us and the next generations.

Honestly, if we aren’t questioning the how and why of the storm that Mythos has been birthing and leaving in its wake, we don’t have the right to call ourselves the smartest species on earth. Because even AI can see where all this is heading.

The question also is: With an Indian as the IMF Head, when are Indian banks going to get their hands on Mythos?

Credits: Featured Image: Gemini
Research assistance for timeline verification: Claude
Thoughts and Writing: Shafali R. Anand