Don’t Be Fooled by This Summer of AI Hype

# Don’t Be Fooled by This Summer of AI Hype The past few months have been a whirlwind of activity, marked by an intense wave of AI hype that has swept across the technology landscape. From bold claims of advanced model capabilities to dramatic security incidents and even purported mathematical breakthroughs, the narrative surrounding artificial intelligence has been dominated by a breathless enthusiasm often amplified by the press. At the close of April, Anthropic ignited headlines by asserting that its Claude Mythos model surpassed most human security experts in identifying software vulnerabilities. This was swiftly followed by the high-profile OpenAI–Hugging Face hacking incident, which prompted Anthropic (with pride) and Meta (with reluctance) to disclose similar security breaches involving their own models. The drama continued as Anthropic then claimed one of its models had achieved a mathematical breakthrough, a feat soon echoed by OpenAI with a similar announcement. Most recently, Anthropic engineer Jacob Coxon went viral upon his departure, warning that both Anthropic and OpenAI are "racing straight towards self-improving superintelligence and gambling with our lives." Each of these events was largely covered uncritically by the media, often repeating the companies’ anthropomorphizing framings. These narratives are carefully crafted to portray their software not merely as powerful tools but as nascent "artificial general intelligence" (AGI), pushing the boundaries of what AI can achieve. This raises a crucial question: Are we truly witnessing a series of massive, civilization-changing technological breakthroughs, or is much of this simply sophisticated marketing? A closer examination reveals a more nuanced picture. In nearly all these incidents, the immense fanfare from the companies—sometimes presented as mea culpas in the case of illicit hacking—is met with intense, often uncritical, press coverage. However, once experts in the relevant fields have the opportunity to scrutinize the claims, a very different story frequently emerges, one that unfortunately garners significantly less media attention. Regarding the "hacking" incidents, cybersecurity experts have largely dismissed the "models gone rogue" or "AI agents creating civilizations" narratives. Instead, they point to OpenAI’s negligence and a fundamental failure to adopt basic, established security practices as the primary culprits. The incidents, they argue, were more a reflection of human error and insufficient safeguards than autonomous AI malevolence. Similarly, the much-touted mathematical results have faced expert skepticism. Mathematicians initially expressed "stunned" reactions to OpenAI’s press release, which claimed its latest chatbot, Astra, had solved problems that "have been open and seen no progress on the main result for at least a decade." Yet, upon deeper review, these experts later concluded that the results were not as "novel as first appeared." The initial shock gave way to a more sober assessment, highlighting that while the AI might have contributed, the breakthroughs were not as groundbreaking or unprecedented as the companies initially suggested. This pattern underscores a critical challenge in the current AI landscape: distinguishing genuine progress from carefully constructed hype. While AI is undoubtedly advancing rapidly, the industry's tendency to anthropomorphize its creations and exaggerate their capabilities often obscures the true nature of these developments, making it imperative for both the press and the public to approach such claims with a healthy dose of skepticism and demand rigorous expert validation.