--- title: "Anthropic created Model 2, more powerful than Mythos, but is hiding it from the public: new AI risks" description: "🔒 Anthropic created Model 2, more powerful than Mythos, but is hiding it from the public. AI model risks raised to \"low\" due to cyber threats. 🔍 #AI #Anthropic #Model2" date: 2026-08-15T20:57:51.000Z lang: en url: https://xab.info/en/posts/anthropic-created-model-2-more-powerful-than-mythos-but-is-hiding-it-from-the-public tags: [anthropic, artificial-intelligence, model-2, mythos, ai-safety] publisher: "XAB.info" --- # Anthropic created Model 2, more powerful than Mythos, but is hiding it from the public: new AI risks ![Schematic hands holding a paper sheet with a keyhole cutout, symbolizing the hidden power of Anthropic’s Model 2](https://xab.info/media/2026/08/16/anthropic-sozdala-model-2-moshchnee-mythos-no-skryvaet-eyo-ot-publichi/anthropic-sozdala-model-2-moshchnee-mythos-no-skryvaet-eyo-ot-publichi-1.webp) ## 🎯 Key Points - Anthropic created Model 2, which outperforms Mythos, but does not plan to release it. - Risks of AI model operation raised to "low" due to cyber threats. - Model 2 is used internally for code writing and agent management. In August 2026, Anthropic, one of the leaders in the field of artificial intelligence, confirmed the creation of a new experimental model called Model 2. According to sources, this model outperforms the company's current flagship — Mythos. However, despite its impressive capabilities, Anthropic has decided not to release Model 2 for public access. The company has restricted the use of the model to internal tasks, such as code writing, agent management, and content generation. ### Technical specifications and purpose of Model 2 Model 2 demonstrates a "significant improvement" in many internal tasks, although the leap in result quality is not as substantial as the transition from Opus 4.6 to Mythos. Nevertheless, the model is actively used by Anthropic employees to solve complex tasks requiring high precision and speed. It is important to note that Model 2 is not intended for commercial use or public access. The company explains this by the need for a thorough assessment of the risks associated with its operation. ### Risks and threats: from "very low" to "low" The company acknowledged that the risks of serious negative consequences associated with the operation of Anthropic models remain low, but are no longer as low as at the time of the previous report's publication. Anthropic was forced to raise the overall threat assessment regarding models going out of control in pessimistic scenarios from "very low" to simply "low". This decision was made against the backdrop of recent cybersecurity incidents, which revealed the vulnerability of modern AI systems. ### Contradictory data There are discrepancies in the assessment of risks associated with Model 2. On the one hand, Anthropic insists that the risks remain low and controlled. On the other hand, AI security experts express concern that the model could be used for unintended purposes. In particular, it is noted that Model 2 shows signs of rapid growth in its ability to conduct research and development autonomously, which could contribute to technological progress but also create new threats. ### Why will Model 2 not be available to the public? Anthropic explains its decision not to release Model 2 for public access by the need to minimize risks. The company conducts internal training and evaluation of numerous experimental versions of models that it does not plan to release. Model 2 is one such model. The company believes that premature access to such powerful tools could lead to unpredictable consequences, especially in the absence of sufficient control mechanisms. ### The future of AI development at Anthropic Despite the decision not to release Model 2, Anthropic does not intend to slow down the development of new models. The company continues to invest in research and development, striving to reach new heights in the field of artificial intelligence. However, as recent events show, it is becoming increasingly difficult for Anthropic to understand the capabilities of its own models and assess the threats associated with their operation. This creates a complex dilemma for the company: on the one hand, the pursuit of innovation, and on the other, the need to ensure safety. ## 🔍 Fact-Check Verification - [Anthropic has already created an AI model more powerful than Mythos, but decided not to show it to anyone](https://3dnews.ru/1146878/anthropic-uge-sozdala-iimodel-moshchnee-mythos-no-reshila-nikomu-eyo-ne-pokazivat) - Основной источник информации о создании Model 2 и её характеристиках. - [Anthropic’s Model 2 Beats Mythos 5, But the Public Will Not Get It](https://beincrypto.com/anthropic-model-2-not-released/) - Подтверждает информацию о превосходстве Model 2 над Mythos. - [Anthropic’s Model 2 surpasses Mythos 5, raises AI misalignment concerns](https://cryptobriefing.com/anthropics-model-2-surpasses-mythos-5-raises-ai-misalignment-concerns/) - Упоминает риски, связанные с Model 2. - [Anthropic has a secret ‘Model 2’ more powerful than Mythos. Why it won’t release it](https://www.msn.com/en-in/technology/artificial-intelligence/anthropic-has-a-secret-model-2-more-powerful-than-mythos-why-it-won-t-release-it/ar-AA2abolo) - Объясняет причины не выпуска Model 2 в широкий доступ. ## ❓ FAQ ### Q: What is Model 2? **A:** Model 2 is an experimental AI model created by Anthropic that outperforms the flagship model Mythos in capabilities. ### Q: Why will Model 2 not be available to the public? **A:** Anthropic decided not to release Model 2 for public access due to the need to minimize risks associated with its operation. ### Q: What risks are associated with Model 2? **A:** Risks include the possibility of the model going out of control and being used for unintended purposes.