The world’s most dangerous AI
This could be the dramatic opening of a sci-fi story, for example:
Treasury Secretary Scott Bessent and Federal Reserve Chair Jerome Powell summoned Wall Street executives to an emergency meeting, fearing that the world’s newest and most advanced AI model could trigger a cybersecurity collapse in the world’s most critical financial IT systems.
Or perhaps:
The new model, developed by Anthropic, one of the world’s leading AI labs, demonstrated capabilities during internal testing that led the company to withhold it from the public for security reasons. The decision was justified by the fact that, in the wrong hands, the new technology poses a threat to the world’s digital infrastructure. Following its predecessors Sonnet (short song) and Opus (masterpiece), the model was named Mythos.
Interestingly, both news items appeared in April on official channels, including Bloomberg’s platforms. At first glance, this paints a decidedly grim picture of the present, but it is worth taking a closer look at what other factors are actually shaping events.
- According to industry reports, Mythos is the largest and, accordingly, most expensive Anthropic model to date. Based on official pricing, it is approximately five times more expensive than the previous top-tier version, Opus 4.6. Meanwhile, it is well known that most AI labs, including Anthropic, operate at a loss across nearly every tier of their subscription packages; even by conservative estimates, they spend several times the monthly fee per user. In this context, making Mythos widely available would not only further increase losses.
- But, due to its significant computational demands, would effectively crowd out the running of other models and slow down further development as well. In a world of finite computational capacity, AI labs must constantly balance inference and training. Capital allocation between short-term inference revenue and long-term model development is one of the most critical strategic decisions, especially when running a model requires five times the capacity of the previous generation. Widespread, subscription-based availability is economically difficult to justify on current infrastructure.
- Mythos Preview truly represents a generational leap in the fields of coding and cybersecurity, yet the picture is more nuanced in the broader benchmark landscape: the three leading frontier models – Mythos, GPT-5.5, and Gemini 3.1 Pro all dominate in different areas, and none clearly stands out above the other two in terms of general intelligence.
- Overall, it can be said that AI model developers are highly capital-intensive companies that constantly require new and increasingly larger rounds of investment. For them, therefore, convincing the next generation of investors is of existential importance, and there are few tools more effective for this than grandiose claims about the capabilities of new models—claims that are difficult to verify.
This does not mean that the development of AI models is not real or significant, but the technological breakthrough and the narrative built around it are two different things, and a significant divergence between the two can distort the markets. When an AI lab withholds a model citing security concerns, it may be a genuine concern, but it could also be a deliberate positioning strategy ahead of the next funding round.
Return Download
This is a marketing communication. Making a well-informed investment decision requires obtaining detailed information. Please read the Key Information Document, the official prospectus, and the management regulations available at the distribution points of the Fund and on the website of the Fund Manager (www.vigam.hu) for detailed information regarding the Fund’s investment policy, distribution costs, and the possible risks of investing. Costs related to the distribution of the investment fund (purchase, holding, sale) can be found in the Fund’s management regulations and at the distribution points. Past performance is not a reliable indicator of future returns. Future returns from the investment may be subject to taxation, and tax and duty information relating to individual financial instruments and transactions can only be accurately assessed based on the individual circumstances of each investor, which may change in the future. It is the investor’s responsibility to obtain information regarding tax obligations.
The data contained in this information material are provided for informational purposes only and do not constitute investment advice, an offer, or investment consulting. VIG Investment Fund Management Hungary Ltd. accepts no liability for investment decisions made based on this information or for their consequences. The license number of the Fund Manager for alternative investment fund management (AIFM) is: H-EN-III-6/2015. The license number of the Fund Manager for UCITS fund management (collective portfolio management) is: H-EN-III-101/2016.