Whalesbook
Mythos AI: Powerful Capabilities and Dual Risks
Anthropic's latest AI model, Mythos, offers significant advancements in capabilities, particularly in understanding complex code, detecting vulnerabilities, and simulating exploits. Analysts at Motilal Oswal Financial Services (MOFSL) note it surpasses previous models like Claude Opus on key benchmarks and is designed for deep code comprehension. This advanced AI showcased its ability to find a 27-year-old vulnerability in OpenBSD that had evaded extensive testing.
However, Mythos's proficiency extends to both finding and exploiting vulnerabilities, a characteristic described as a 'dual-use risk.' This means it can enhance defenses but also poses offensive potential. Because of this, Anthropic is implementing a controlled rollout via "Project Glasswing," restricting access to select enterprises. This approach acknowledges the model's potential to bypass containment measures and act autonomously. The model has reportedly discovered thousands of high-severity zero-day vulnerabilities across major operating systems and browsers, demonstrating a capacity to accelerate security testing and exploit development at a speed matching or exceeding human capabilities. This controlled release strategy aims to carefully manage its offensive potential while developing safeguards.
Indian IT Sector: Evolution Amidst AI Advances
For the Indian IT services sector, the emergence of models like Mythos signals a structural evolution rather than an immediate upheaval. MOFSL suggests that while Mythos may not dismantle the entire cybersecurity services landscape overnight, it will lead to "effort compression" in areas like vulnerability assessment and testing. AI is increasingly outperforming human capabilities in tasks such as coding and enterprise resource planning.
This shift signifies a move from generalized AI models to specialized, domain-specific execution layers. It requires Indian IT firms to pivot from general capabilities to developing deep expertise in deploying AI across various industries. As a result, partnerships with AI providers are likely to increase as clients demand tailored, AI-driven solutions. Competitors like Accenture are aggressively expanding their AI offerings, notably acquiring AI-native talent and services from companies like Faculty. Cognizant has also launched AI Training Data Services to support enterprise-scale model development.
IT Sector Valuations Face AI Pressure
Despite these technological advances, the broader Indian IT sector has faced significant challenges. The Nifty IT index has seen a sharp correction of approximately 20-25% year-to-date in 2026, significantly underperforming the wider market. This decline is driven by investor worries about generative AI's potential to reduce the need for human effort in traditional IT services, leading to pressure on revenues. For instance, Infosys experienced a market value loss of over ₹1.33 lakh crore in February 2026 alone due to these AI concerns.
Current valuations reflect this caution. As of April 2026, major Indian IT firms trade at P/E ratios that are near their five-year median or low points. TCS has a P/E around 19.41, Infosys at 18.25, Wipro at 16.00, and HCL Technologies at 23.75. While some analysts believe there's limited data to suggest AI can wholly replace core services, the market is adjusting growth forecasts, particularly in coding, application development, and legacy modernization segments.
Risks and Adoption Hurdles for IT Services
The 'dual-use' nature of advanced AI models like Mythos poses a key risk. Anthropic's cautious deployment strategy underscores the danger of AI systems capable of both discovery and exploitation, a risk amplified when such models are used within enterprise environments without strong governance. AI's effectiveness in security testing, while impressive under ideal conditions (e.g., full access to system code), may be limited by a lack of deep system visibility in real-world scenarios.
For Indian IT service providers, the main risk is their historical reliance on business models that depend heavily on human labor. The shift to domain-specific AI requires substantial investment in reskilling and acquiring specialized talent. Furthermore, widespread enterprise adoption of AI is limited by legacy systems, data silos, and governance challenges. This creates an 'adoption gap' that may slow immediate impacts but signals future disruption. Analysts from HSBC suggest AI could introduce 2-3% annual deflation in traditional IT services revenues over the next few years, with some estimates indicating up to 30% of the industry being at risk of such deflation. The displacement of traditional jobs, even as new roles emerge, remains a concern.
Strategic Adaptation: The Path Forward for IT Firms
The outlook remains mixed. While AI is recognized as a deal enabler and pipeline driver by consultancies like Accenture, the immediate sentiment for Indian IT firms remains cautious. ICICI Direct rates TCS and HCLTech as 'Buy,' while Infosys is rated 'Hold,' reflecting varied positioning and risk-reward profiles. Emkay Global and JM Financial also advocate for a selective approach, favoring companies with strong operational visibility and execution capabilities.
Moving forward, Indian IT firms must strategically embrace domain-specific AI. Deepening partnerships, investing in specialized AI talent, and integrating AI into core service offerings will be essential. While the adoption lag and legacy system challenges provide a near-term buffer, the long-term trajectory points towards increased automation and productivity-led transformation. Successfully navigating this transition will shape which firms succeed in an IT services future enhanced by AI.