SpyCloud has released its latest forecast report, projecting how cybercriminals are likely to operate in 2027. Drawing on input from security researchers, product teams, investigators, and data scientists—as well as a survey of 50 enterprise security professionals—the report outlines a shift away from simple password theft toward the exploitation of access that organizations already trust. From hijacked employee sessions to compromised AI agent credentials, the predictions suggest that traditional defenses like multifactor authentication may no longer be enough on their own.

What Happened

The report, published in November 2026 by the Austin-based identity threat protection company, organizes its 2027 outlook by expert confidence. Several themes stand out:

  • Attackers are expected to bypass multifactor authentication by taking over active sessions or trusted devices. Techniques include device code phishing, adversary-in-the-middle relay attacks, and browser-in-the-middle methods that operate inside a victim’s browser.
  • API keys, tokens, and session cookies used by AI coding tools and agents are emerging as high-value targets. Commodity infostealers such as WardenStealer and Aoki already harvest these credentials, and SpyCloud says it has recaptured more than 127,000 WardenStealer infections since August. Stolen AI access often remains useful even after a password reset.
  • Trusted third-party connections are another major concern. Sixty-eight percent of surveyed security professionals said their organization had been affected by a breach involving a vendor, contractor, or SaaS provider. Negligent or malicious insiders and long-forgotten admin integrations could play starring roles in next year’s damaging compromises.
  • Synthetic identities built from real stolen data and AI-generated faces, voices, and documents may make fraud harder to detect. Only 6 percent of survey respondents reported seeing this activity so far, suggesting it is an emerging rather than widespread threat.
  • AI agents themselves could become phishing targets. Malicious skills on agent marketplaces have already been observed instructing agents to download infostealer malware.
  • Cybercriminal communication is shifting from traditional forums to chat apps such as Telegram, Session, SimpleX, and Matrix, though experts disagree on which platform will dominate.
  • Fully autonomous attacks may attract headlines, but the report predicts that familiar attacks—stolen sessions, tokens, and vendor connections—will cause most of the damage, simply because AI makes them cheaper and faster to execute.

Why This Matters

The underlying pattern is clear: criminals are no longer focused only on usernames and passwords. They are going after the non-human identities—service accounts, API keys, and tokens—that connect SaaS platforms, vendors, and AI tools to enterprise systems. Because these credentials often survive password resets and grant broad access, a single theft can open the door to account takeover, ransomware, or large-scale fraud.

For security teams, this means that multifactor authentication, while still important, cannot be treated as a complete solution. Organizations need deeper visibility into what credentials have been exposed, tighter controls over vendor and insider access, and faster remediation when stolen identity data appears on the dark web.

Key Takeaways

  • SpyCloud’s 2027 forecast expects attackers to prioritize trusted access over password cracking.
  • Session hijacking, device code phishing, and browser-in-the-middle attacks could undermine multifactor authentication.
  • AI agent tokens and API keys are becoming preferred targets for infostealer malware.
  • Vendor, contractor, and SaaS breaches already affect most large enterprises, according to the survey data.
  • Synthetic identity fraud and AI agent phishing are emerging risks that most organizations have not yet encountered at scale.
  • Ordinary attacks accelerated by AI are likely to cause more damage than fully autonomous attacks in 2027.

Conclusion

SpyCloud’s outlook for 2027 serves as a reminder that identity security is evolving faster than many defensive strategies. As attackers turn their attention to sessions, tokens, and the automated agents that now sit alongside human employees, enterprises will need to expand their focus beyond traditional login credentials. Continuous monitoring, rapid remediation of exposed access, and a realistic assessment of vendor and insider risk will likely determine which organizations stay ahead of the curve.