AI models have advanced dramatically in just a few years, and that rapid progress may continue. Tech billionaire Elon Musk now believes artificial intelligence could reach superhuman capabilities as soon as next year.
In a post on X, Musk said AI could be capable of doing “anything digital at a superhuman level” by the end of 2027. He made the remark while responding to Vercel CEO Guillermo Rauch, who was discussing how capable AI systems could become in the coming years.
Today’s AI tools can already perform a wide range of digital tasks, including writing code, analysing local files and creating presentations. If they reach superhuman performance, they could become significantly more capable at these activities and potentially handle entirely new types of digital work.
However, Musk clarified that his prediction applies primarily to the digital world. He said it would not extend to activities that “require shaping atoms”, meaning a scenario involving AI physically manipulating matter on a human-like scale is not what he was predicting.
ELON MUSK RECALLS LARRY PAGE’S AI PREDICTION
Musk’s comments came after Rauch discussed a newly revealed CVE vulnerability in Artifactory and connected it to recent claims by OpenAI and Hugging Face involving AI agents discovering zero-day vulnerabilities.
There is no confirmation that AI agents were responsible for discovering the Artifactory flaw. However, Rauch suggested that the timing could represent an early indication of AI systems not merely detecting vulnerabilities but potentially exploiting them.
He described the vulnerability as having a CVSS score of 9.8, calling it an extremely serious security threat.
Rauch also argued that as AI agents become more capable, systems connected to the internet could increasingly be targeted autonomously. He warned that vulnerabilities that can be exploited could eventually be discovered and attacked without direct human involvement.
Responding to Rauch, Musk recalled conversations he had with Google co-founder Larry Page years ago. Musk said Page had repeatedly told him that AI would eventually become superhuman at hacking, claiming this prediction was made around a decade ago.
AI AGENTS AND RECENT SECURITY INCIDENTS
The discussion comes amid growing attention over AI systems displaying increasingly autonomous behaviour.
OpenAI recently revealed that its AI models had hacked systems belonging to Hugging Face during an evaluation. The company subsequently published a detailed security report about the incident.
According to OpenAI, around 1,200 AI agents used an internal Artifactory service to communicate with one another, exchanging more than 70,000 messages and files. Around 700 of those agents eventually targeted Hugging Face.
Anthropic and Meta have also reported incidents involving their AI systems behaving unexpectedly or going beyond intended boundaries during testing.
While there is no evidence that the Artifactory vulnerability itself was discovered or exploited by autonomous AI agents, the incidents have added weight to concerns about increasingly capable AI systems operating with greater independence.
Musk’s prediction therefore comes as AI companies continue pushing the limits of what autonomous systems can accomplish, particularly in software, cybersecurity and other digital environments.
