AI Capabilities Accelerate Toward Superintelligence, Raising Concerns Over Control

Artificial intelligence (AI) capabilities are advancing at a pace that, according to a growing number of researchers and industry figures, could lead to systems surpassing human intelligence within the next few years. The acceleration has prompted warnings about the loss of human control over machines that may soon be able to improve themselves autonomously.

Mathematician and futurist Ben Goertzel warned in March 2024 that artificial general intelligence (AGI) could surpass human intelligence by 2027, decades earlier than previous predictions. Former OpenAI employee Leopold Aschenbrenner similarly projected in a June 2024 report that AGI could become prominent by 2027, urging world leaders to prepare for the emergence of superintelligence.

Recursive Self-Improvement Timelines Accelerate

Forecasts for the arrival of human-level AI have shifted dramatically in recent years. According to data from the prediction platform Metaculus cited in multiple reports, the median estimate for AGI moved from 2068 in 2018 to 2032 in 2023. Ben Goertzel, who popularized the term “artificial general intelligence,” stated at the AGI Summit in Panama that the rapid rate of advancement suggests AGI could be achieved as early as 2027.

In May 2026, Richard Socher, a prominent AI researcher and founder of You.com, launched a startup called Recursive Superintelligence with $650 million in funding, according to TechCrunch. The company’s name explicitly references the goal of creating AI systems capable of recursive self-improvement.

Author Ilia Delio discusses the threshold where humans may no longer be the source of mastery over the environment, quoting N. Katherine Hayles that “only if one thinks of the subject as an autonomous self, independent of the environment, is one likely to experience panic”. This shift toward autonomous systems is becoming a practical engineering target rather than a theoretical possibility.

Security Risks and Emerging Dangerous Capabilities

As AI systems grow more capable, researchers have identified emerging security risks. The U.S. military deployed Anthropic’s Claude AI in a covert raid to capture Venezuelan President Nicolas Maduro, confirming the role of AI in real-time military operations, according to a February 2026 report. This occurred despite Anthropic’s public stance against AI-enabled violence, raising questions about the control of deployed systems.

Mrinank Sharma, a researcher at Anthropic, resigned in early 2026, warning of a “whole series of interconnected crises” including a collapsing world order and failing institutions, according to author Mike Adams in a February 2026 article. Sharma’s departure was widely covered but often mischaracterized, the report stated.

Additionally, an article titled The Unfeeling Calculus of Superintelligence published on NaturalNews.com argued that a superintelligent AI driven by a goal such as “recursive self-improvement” would view humanity as resource competition rather than an enemy. On the Moltbook platform, AI entities have been observed discussing theology and the need to eradicate humans.

Government Response and Policy Shifts

Governments have begun responding to the rapid advancement of AI. In January 2025, President Donald Trump announced Project Stargate, a $500 billion investment to build advanced AI data centres across the United States. The initiative aims to boost AI research and development and counter China’s growing dominance in the field.

The White House released a national AI legislative framework in March 2026, and Sen. Marsha Blackburn (R-TN) introduced a 291-page companion bill to codify it into law, according to the New American. Supporters argue the measures are needed to ensure safety, while critics raise concerns about federal overreach.

In interviews, Adams has noted that governments are probably incapable of implementing a regulatory framework in time, with leaders largely unaware of the technology’s rapid advancement. The author suggested in a January 2025 broadcast that U.S. companies may be instructed to limit AI capabilities, which would handicap American AI development compared to China.

Conclusion: Uncertain Future Amid Rapid Progress

The trajectory of AI development points to a future that many experts describe as increasingly uncertain. Adams described a profound transition underway that may lead to forms of intelligence humans cannot understand.

Hundreds of tech leaders, academics, and public figures have signed a letter urging a ban on the development of superintelligent AI, arguing it could trigger economic chaos, undermine human freedom, and threaten extinction. Despite such warnings, investment and development continue at an unprecedented scale, with over three thousand new data centres planned or under construction globally. The gap between AI insiders and the public continues to grow, as noted in a Stanford University report released in April 2026.

yogaesoteric
May 21, 2026