AI Hype Masks Marketing and Security Lapses, Experts Say

The summer of 2024 has been characterized by significant hype surrounding artificial intelligence, with major companies like Anthropic and OpenAI making bold claims about their models' capabilities and security incidents. These announcements, often amplified by breathless press coverage, frequently frame the AI advancements as steps towards artificial general intelligence (AGI). However, a closer examination by cybersecurity and mathematics experts suggests that much of this fanfare is driven by marketing rather than genuine, unprecedented technological leaps, and that security breaches are often attributed to basic negligence.

At the end of April, Anthropic announced that its model, Claude Mythos, demonstrated superior ability in identifying software vulnerabilities compared to most security professionals. This was followed by a period of disclosed hacking incidents involving AI models. Both Anthropic and Meta reported such breaches, with OpenAI also experiencing a significant incident. Shortly after these security disclosures, Anthropic claimed one of its models had achieved a mathematical breakthrough. OpenAI quickly followed with its own announcement of a mathematical discovery made by its chatbot, Astra. More recently, Anthropic engineer Jacob Coxon gained viral attention by announcing his departure, stating that both Anthropic and OpenAI are "racing straight towards self-improving superintelligence and gambling with our lives."

While these events were widely reported with significant fanfare, often repeating the companies' anthropomorphizing language, independent expert analysis has presented a different narrative. Cybersecurity experts, in response to the "hacking" incidents, have indicated that the issues stem more from OpenAI's "negligence and failure to adopt basic, established security practices" rather than from "models gone rogue" or "AI agents creating civilizations." Similarly, mathematicians who were initially "stunned" by OpenAI's press release regarding Astra solving long-standing mathematical problems found that the results were "not as novel as first appeared" upon closer inspection. This pattern suggests that the intense media attention and company pronouncements often overshadow a more nuanced reality where marketing plays a significant role in shaping public perception of AI advancements, and where fundamental security and scientific rigor are sometimes overlooked in the race for public attention.

The trend of companies presenting incremental progress or security lapses as major breakthroughs highlights a broader concern within the AI community regarding transparency and the responsible communication of AI capabilities. The emphasis on AGI and the dramatic framing of discoveries can obscure the actual state of the technology and the practical challenges involved in its development and deployment. Experts suggest that a more measured and evidence-based approach to reporting AI progress would benefit both the public understanding of AI and the long-term credibility of the field. The contrast between the initial breathless reporting and the subsequent expert deconstructions underscores the need for critical evaluation of AI claims, particularly when they are accompanied by significant marketing efforts and emotional appeals.