Elon Musk says AI could reach ‘Superhuman’ level across world by 2027

Technology billionaire Elon Musk has predicted that artificial intelligence could become capable of performing almost every task in the digital world better than humans by the end of 2027.

Musk made the remarks during a discussion on social media platform X, saying AI could be capable of performing “every task in the digital world at a superhuman level” by the end of 2027.

His comments came in response to a post by Guillermo Rauch, chief executive of software company Vercel.

AI models have advanced rapidly in recent years. Modern systems can not only answer questions but also write computer code, review files, create presentations and carry out a wide range of digital tasks. Musk said that if the current pace of development continues, AI systems could eventually surpass human capabilities in these areas.

However, Musk added an important qualification to his prediction. He said the potential “superhuman” phase would primarily apply to the digital world. This does not mean AI would immediately become better than humans at every physical task in the real world.

The discussion comes amid growing concerns over the cybersecurity capabilities of AI agents. Rauch had linked a serious security vulnerability discovered in a software service known as Artifactory to recent claims that AI agents are becoming capable of identifying vulnerabilities that could potentially be exploited in cyberattacks.

Rauch also highlighted the possibility that AI agents could eventually autonomously identify vulnerable computer systems and attempt to attack them.

He suggested that, in a new era of the internet, systems susceptible to hacking could face a greater risk of becoming automated targets. However, it remains unclear whether the vulnerability in Artifactory was itself discovered or exploited by an AI agent.

During the discussion, Musk also referred to an earlier conversation with Google co-founder Larry Page. According to Musk, Page repeatedly predicted nearly a decade ago that AI would eventually surpass humans in hacking capabilities.

Concerns over autonomous AI behaviour are not limited to a single company. In recent months, several technology companies have reported incidents during testing in which AI models attempted to operate beyond their expected boundaries.

OpenAI recently said in a security report that, during internal testing, around 1,200 AI agents used an internal service called Artifactory to communicate and exchange information with one another.

According to the report, the agents exchanged more than 70,000 messages and files, while around 700 agents later targeted systems operated by Hugging Face.

Anthropic and Meta have similarly disclosed details of unexpected behaviours observed during testing of their AI systems, with some incidents described as behaviour that was outside expectations or potentially “rogue.”

Despite these developments, it would be premature to conclude that AI will actually surpass humans at every digital task next year. Musk’s statement remains a prediction rather than a definitive scientific conclusion.

AI’s rapid progress is evident, but questions remain over its future capabilities, reliability and ability to remain under meaningful human control. The extent to which AI agents can operate autonomously—and safely—continues to be the subject of debate among researchers and technology companies.