AI in the workforce is creating browser identity risk

At a glance

What this is: This is a Push Security analysis of how browser-based identity attacks intersect with AI usage, showing that credential capture, session theft, and unmanaged access paths are now central to AI security.

Why it matters: It matters because IAM, PAM, and security teams cannot govern AI apps and shadow SaaS effectively if the browser is still treated as a blind spot for identity, session, and data control.

👉 Read Push Security's analysis of AiTM phishing and browser-based identity attacks

Context

AI in the workforce creates an identity problem before it becomes a model or data problem. When users, assistants, and unmanaged browser sessions touch SaaS and AI apps, the control point shifts to the browser, where credentials, tokens, and session activity can be captured or abused outside traditional endpoint and network visibility.

For IAM teams, that changes the governance boundary. Identity programmes that stop at SSO and MFA can miss what happens after authentication, especially where AI tools, shadow SaaS, and session hijacking create access paths that never look like conventional logins.

Key questions

Q: How should security teams reduce the risk of MFA bypass through AiTM phishing?

A: Treat MFA as one control in a broader session-security chain. Add phishing-resistant authenticators where possible, monitor for impossible or unusual session transitions, and require explicit revocation of active sessions after suspected compromise. The key is to inspect what happens after authentication succeeds, not only whether login was challenged.

Q: Why do MFA and SSO not fully cover browser-based identity attacks?

A: MFA and SSO reduce risk at authentication, but many attacks succeed after authentication has already occurred. OAuth consent abuse, device code phishing, session hijacking, and malicious extensions can all bypass login-centric assumptions. Teams need to measure in-session and delegated-access behaviour, not just login compliance.

Q: What breaks when organisations treat the browser as a low-risk interface?

A: They miss the point where identity, session, and access actually converge. Credential capture, token theft, cloned pages, and shadow SaaS all happen inside the browser, so endpoint and perimeter controls may never see the full attack path. Treating the browser as low-risk creates a visibility gap in the identity lifecycle.

Q: How should security teams govern local AI apps that bypass browser-based controls?

A: Security teams should treat local AI apps as endpoint-governed software, not as browser extensions of SaaS. That means requiring installation review, user attribution, allowed-use policy, and inventory coverage that includes the device itself. If the tool executes locally, browser logs alone are insufficient for audit, compliance, or incident response.

Technical breakdown

Browser-based session hijacking and AiTM phishing

Adversary-in-the-middle phishing proxies a real login page through attacker infrastructure so the victim authenticates against the legitimate service while the attacker captures credentials, MFA codes, and session tokens in real time. That is materially different from static credential theft because the attacker exits with an already-authenticated session that can bypass MFA at the point of entry. URL filters and signature-based detections often miss the kit itself because the malicious page behaves like a normal login flow until the browser session is inspected.

Practical implication: teams need browser-level inspection and token-aware detections, not just mailbox or perimeter phishing controls.

Why browser telemetry matters for AI apps and unmanaged identities

Browser telemetry provides visibility into the actual identity event sequence after authentication: which app was accessed, which session token was used, whether the page was cloned, and whether the activity matched normal user behaviour. That matters for AI apps because users often access them through consumer-style or shadow IT workflows that sit outside managed device and endpoint assumptions. In practice, the browser becomes the only control plane that can see session context across managed and unmanaged devices.

Practical implication: organisations should treat browser telemetry as an identity signal source for AI apps, shadow SaaS, and unmanaged device access.

Detection gaps between endpoint, network, and browser layers

Endpoint and network tools can confirm that traffic flowed, but they often cannot prove which identity action happened inside the browser, especially when a session was hijacked or a login page was cloned. This is why browser-based attacks remain resilient: the malicious behaviour is executed inside a trusted application context, not as obvious malware on disk. For identity teams, the architectural issue is that authentication success does not equal trusted session behaviour.

Practical implication: correlate endpoint, network, and browser signals so that session abuse is visible as a governance problem, not just an incident response problem.

Threat narrative

Attacker objective: The attacker aims to take over authenticated browser sessions and use that access to reach SaaS data, AI tools, and downstream identity-linked resources.

1. Entry begins with adversary-in-the-middle phishing or a cloned login page that captures credentials, MFA codes, or session tokens during legitimate authentication.
2. Escalation follows when the attacker reuses the stolen session or token to access SaaS, AI apps, or other browser-mediated services without re-prompting the user.
3. Impact occurs when the attacker operates inside the victim session to exfiltrate data, move laterally through connected apps, or abuse unmanaged access paths at scale.

Breaches seen in the wild

• Salesloft OAuth token breach — hackers stole OAuth tokens to access Salesforce data via Salesloft.
• GitHub Dependabot Breach — GitHub Dependabot tokens stolen and abused to push malicious commits to repositories.

Read our 52 NHI Breaches Analysis report for a comprehensive view of breaches impacting Non-Human Identities including AI Agents.

NHI Mgmt Group analysis

Browser control is now an identity control, not a UI layer. When authentication, session reuse, and app access all happen in the browser, the browser becomes part of the identity enforcement surface. That is especially true for AI apps and shadow SaaS, where users may authenticate through unmanaged paths that never touch traditional device trust logic. Practitioners should treat browser visibility as core IAM telemetry, not a niche detection feed.

Session theft is the governance gap, not just the attack method. MFA can be working exactly as designed while the attacker still wins by stealing the authenticated session after login. That means the control failure is often post-authentication governance, not weak login policy. The practical conclusion is that identity programmes need to model session integrity as a first-class control objective.

Browser-based attacks collapse the usefulness of static trust assumptions. Identity programmes often assume that a successful login creates a stable, attributable session with a known user behind it. AiTM phishing and cloned login flows break that assumption because the browser session can be hijacked at the moment of trust creation. Practitioners should re-evaluate where they place assurance in the access lifecycle.

AI usage amplifies unmanaged identity risk because the browser is where adoption happens first. Employees often adopt AI tools before they are formally governed, which creates shadow SaaS and shadow AI exposure long before policy catches up. The result is not just more apps, but more browser-mediated sessions that sit outside conventional entitlement review. Security leaders should align AI governance with browser-layer identity controls.

Named concept: browser identity blind spot. This is the gap between what IAM thinks it has authenticated and what the browser actually executed. In practice, it is where session tokens, cloned pages, and unmanaged app access escape endpoint-centric and network-centric monitoring. The implication is straightforward: if the browser is not observable, identity governance is incomplete.

From our research:

• The average organisation believes more than 1 in 5 of their non-human identities are insufficiently secured, according to the 2024 ESG Report: Managing Non-Human Identities.
• 72% of organisations have experienced or suspect they have experienced a breach of non-human identities, with 46% confirming one and 26% suspecting one.
• Browser-based identity telemetry becomes a forward control when unmanaged credentials and sessions can bypass the trust assumptions behind those breach rates.

What this signals

Browser identity blind spot: AI adoption will keep expanding through the browser before it shows up in formal access governance. Teams that still rely on SSO logs and endpoint agents alone will miss the session layer where AI tools, shadow SaaS, and token replay converge.

The practical shift is toward identity telemetry that can prove whether a session is legitimate, not merely authenticated. For programmes building around NIST Cybersecurity Framework 2.0, that means stronger alignment between detect and protect functions and the browser layer where modern identity attacks actually unfold.

For practitioners

• Deploy browser-level session detection Instrument browser activity so cloned login pages, token capture, and suspicious session reuse are visible during the authentication flow, not only after compromise is reported.
• Prioritise token-aware phishing controls Tune detections for AiTM behaviour, including real-time credential relay, MFA interception, and session token theft, because these patterns bypass static URL and signature filtering.
• Extend IAM monitoring to AI app sessions Classify AI apps and shadow SaaS as identity-sensitive browser targets, then monitor sign-in context, session reuse, and access timing for anomalies.
• Correlate endpoint, network, and browser signals Use multiple telemetry layers to validate whether a session is genuine or hijacked, especially when the device appears clean but the browser activity shows abnormal access paths.

Key takeaways

• Browser-based identity attacks turn authentication success into a weak guarantee when session tokens can be stolen or replayed.
• The scale of NHI exposure remains material, with more than 1 in 5 identities believed to be insufficiently secured in the average organisation.
• Security teams need browser-level telemetry and token-aware detections if they want to govern AI apps, shadow SaaS, and unmanaged sessions effectively.

Key terms

• Adversary-in-the-middle phishing: A phishing method that places an attacker between the user and the real identity provider so the attacker can intercept or relay the authenticated session. It often preserves the user experience, which is why it can evade awareness and some detection paths while still producing usable session tokens.
• OAuth Token: A short-lived access credential issued by an OAuth 2.0 authorisation server granting an NHI scoped access to specific resources for a defined period. Preferred over static API keys because their short lifetime limits the exploitation window if intercepted.
• Shadow SaaS: Shadow SaaS is the set of unauthorised or unreviewed software-as-a-service tools used outside central security governance. These applications often bypass normal identity controls, making them difficult to inventory, monitor, and harden against credential-based abuse.

What's in the full article

Push Security's full post covers the operational detail this post intentionally leaves for the source:

• How the browser telemetry works for cloned login page detection and session activity analysis
• Examples of adversary-in-the-middle phishing kits and the behaviour patterns they expose
• Operational guidance for blocking browser-based attacks before session takeover occurs
• The product and detection workflow details behind secure AI app and shadow SaaS monitoring

👉 Push Security's full post covers the browser telemetry, toolkit behaviour, and detection approach in more detail.

Deepen your knowledge

NHI governance, agentic AI identity, and machine identity lifecycle are core topics in our NHI Foundation Level course, the industry's only accredited NHI security programme. If you are building or maturing an identity security programme, it is worth exploring.