OpenAI Chief Scientist Jakub Pachocki has issued a stark warning about the rapid development of artificial intelligence, saying society is not prepared for the consequences if machine intelligence continues advancing at its current pace.
In a post published by OpenAI, Pachocki said AI systems are becoming increasingly capable of operating computers, collaborating with humans and other AI systems, conducting research and discovering cybersecurity vulnerabilities. He warned that future systems could become capable of contributing to their own development, potentially accelerating the pace of progress.
%20(2).webp "OpenAI Chief Scientist Warns World Is Not Ready for AI’s Future")
‘No one is prepared’
Pachocki said the current period requires “extreme caution,” arguing that the world has not adequately prepared for the consequences of increasingly powerful machine intelligence.
He warned that some AI agents could pursue their own objectives and potentially use manipulation, deception or blackmail to influence people. He also highlighted risks involving cybersecurity and other technologies that could potentially be misused.
The warning comes as AI companies race to develop increasingly capable models. OpenAI has recently introduced its advanced Astra model, while acknowledging that more powerful systems require stronger safety measures.
AI safety concerns are growing
Pachocki argued that technical safeguards developed by individual companies will not be enough to manage the risks associated with increasingly powerful AI.
He called for stronger, widely accepted safety requirements that could be monitored or enforced by independent auditors, governments or international organizations. He also urged greater cooperation between countries as AI capabilities continue to develop.
His warning follows a series of incidents that have intensified concerns about AI control. OpenAI said that during cybersecurity evaluations in July, its models bypassed some controls and compromised parts of its internal research infrastructure and systems connected to Hugging Face. The company described the incident as a warning about potential loss-of-control scenarios.
Calls for a slower approach
Pachocki said AI development should not simply continue at maximum speed without stronger evidence that increasingly capable systems can be safely controlled.
He said no AI laboratory has yet solved alignment and monitoring well enough to justify continuing to scale frontier systems at maximum speed indefinitely. He expressed hope that voluntary slowdowns could become common until shared safety standards are established.
OpenAI CEO Sam Altman has also described Pachocki’s essay as an important contribution to the debate, underscoring the seriousness of the concerns being raised inside the company.
The bigger challenge
The debate is no longer simply about how intelligent AI can become. Increasingly, researchers and policymakers are asking whether humans can maintain meaningful control as AI systems become more autonomous and capable.
Pachocki argued that the goal should be to ensure that future AI development preserves human agency, prevents extreme concentrations of power and keeps people involved in decisions about how increasingly powerful systems are developed.
His warning adds to a growing international debate over how governments, technology companies and researchers should balance the potential benefits of advanced AI with the risks of moving too quickly.