Deloitte: Companies are deploying AI agents faster than safety measures are keeping up

Businesses are adopting AI agents faster than they are building safety protocols, Deloitte said in its State of AI in the Enterprise report, based on a survey of more than 3,200 business leaders across 24 countries. The report found 23% of companies are currently using AI agents "at least moderately," a figure projected to rise to 74% within two years, while the share not using them at all is expected to fall from 25% to 5%.

Deloitte warned the rise of agentic AI — tools trained to perform multistep tasks with little human supervision — is not being matched by adequate guardrails. Only about 21% of respondents said their company currently has robust safety and oversight mechanisms in place. The report noted greater agent autonomy increases risk because agents can interact with systems to sign documents or make purchases and are vulnerable to prompt injection and unexpected behaviour, sometimes with "disastrous consequences." Other recent studies cited in the article echo the divide between deployment and safety.

One May study found 84% of IT professionals said their employers were already using AI agents while just 44% said they had policies to regulate those systems. A September report from the nonprofit National Cybersecurity Alliance said many people use AI tools at work without employer safety training, and a December Gallup poll showed 23% of workers did not know if their employers were using AI at the organizational level.

Key Topics

Tech, Deloitte, Ai Agents, Safety Protocols, Prompt Injection, Openai