BeyondTrust outlines cybersecurity predictions for 2027 and beyond

0

BeyondTrust has released a set of cybersecurity predictions for 2027 and beyond, arguing that autonomous AI agents, faster exploitation and tighter oversight will reshape how organisations manage identity, privilege and governance.

The forecast, developed by BeyondTrust executives, security experts and researchers from BeyondTrust Phantom Labs, focuses on risks linked to non-human identities and “machine-speed” cyberattacks, as well as increasing regulatory and board scrutiny of AI controls.

“As AI agents take on more responsibility across the enterprise, the challenge is no longer simply determining who or what has access, but understanding what those identities are authorised to do, how far their authority extends, and who remains accountable for their actions,” said Janine Seebeck, Chief Executive Officer at BeyondTrust. “The organisations that succeed in this next era of AI will be those that build security and accountability into how autonomous systems operate, maintaining control over privilege without limiting the innovation these technologies make possible.”

BeyondTrust’s five headline predictions include what it describes as the first major AI agent-related incident being handled like an insider threat, with organisations needing clear authority to shut down autonomous systems. Another prediction centres on “over-privileged” AI agents becoming a primary attack path if adversaries exploit legitimate permissions to access sensitive systems, reinforcing least-privilege approaches.

The company also points to AI-assisted vulnerability research accelerating discovery, particularly in open-source software, and said the resulting volume could make “patch everything” approaches impractical, increasing the importance of threat-intelligence-led prioritisation. It further predicts the possibility of a publicly acknowledged end-to-end autonomous breach in 2027, where AI agents identify vulnerabilities, chain exploits and exfiltrate data while AI-enabled defenders respond at machine speed.

On governance, BeyondTrust predicts that regulators, auditors, insurers and customers will increasingly demand evidence of effective AI control frameworks, with potential financial and regulatory consequences for organisations unable to demonstrate adequate oversight.

BeyondTrust also referenced recent Phantom Labs research into AI coding-agent activity on enterprise endpoints. It said a four-week analysis observed approximately 1.28 million shell commands requested from 132,000 AI coding-agent prompts, averaging nearly 10 commands per prompt. The research also found 85% of observed coding-agent users had at least one Model Context Protocol server configured.

The company said the findings underscore the need for organisations to reconsider identity security and privilege governance as AI agents expand their interactions across connected systems.

Share.