Loading story…
Cloudflare has introduced optional OAuth scopes for third-party applications, replacing all-or-nothing consent with task-based authorization. OAuth client owners can designate requested permissions as required or optional, enabling users to deselect individual optional scopes; tokens issued after authorization contain only the permissions the user approved. The feature is particularly intended to limit overprivileged access in AI-agent and Model Context Protocol (MCP) integrations, which may otherwise request permissions for every potential action.
The change applies only to scopes included in a given authorization request, and existing OAuth clients retain all-or-nothing behavior unless their owners opt in. Applications using optional scopes must inspect the scopes actually granted after authorization and safely reduce functionality where users decline permissions, rather than assuming a successful authorization supplied every requested privilege. Cloudflare reports that developers have created thousands of third-party OAuth apps and more than one million authorizations since launching its third-party app capability, and plans to extend role and scope coverage across most Cloudflare products.
See affected versions and whether adversaries are exploiting it.
2 events from the most recent confirmed update back to the earliest known activity.
The MCP 2026-07-28 specification preferred pre-registered clients and Client ID Metadata Documents while deprecating Dynamic Client Registration.
Cloudflare added OAuth scope customization that lets third-party OAuth client owners mark scopes as optional, allowing users to decline those permissions during consent. Tokens issued after consent contain only the scopes the user granted, and applications must inspect the granted scope set and handle partial access.
Vulnerabilities, threat actors, malware, products, organizations, and breaches Mallory has linked to this story.
See whether adversaries are exploiting this yet, and where the affected versions run in your environment.
3 references tracked. Mallory keeps watching after this page renders.
infoq.com
Open sourceblog.cloudflare.com
Open sourcelearn.microsoft.com
Open sourceMap indicators from this story to your assets and identify affected systems in minutes.
Every observed campaign, victim, and pivot linked to actors named in this story.
Malware, exploits, and IOCs connected to the activity described here.
YARA, Sigma, and Snort rules deployed to your SIEM as soon as they’re published.
Get matching new stories delivered to your team as they break — not the next morning.
Ask questions about this story and take action on the answers.
Other stories Mallory is tracking that overlap on entities, sources, or topic labels.
Security researchers have identified a new phishing technique called **ConsentFix** that enables attackers to compromise cloud accounts by exploiting legitimate OAuth consent flows. Instead of stealing passwords or MFA codes, the attack relies on social engineering to trick users into copying and pasting a valid Microsoft sign-in URL, which contains an OAuth authorization code, back into a phishing page. This allows threat actors to capture OAuth authorization codes and access tokens, granting them full access to the victim's cloud account without credential exposure. The attack specifically targets users by requiring a matching email address and leverages trusted authentication infrastructure, making detection more challenging for organizations relying on cloud services. The ConsentFix method is similar to the previously documented ClickFix technique, which used fake CAPTCHA pages, but it innovates by harvesting tokens directly through the OAuth process. The phishing flow involves a Cloudflare-style verification prompt, targeted email validation, and redirection to a legitimate Microsoft login page. After successful authentication, the victim is instructed to paste a localhost URL containing the authorization code into the phishing site, enabling the attacker to hijack the session. This technique poses a significant risk to organizations using cloud platforms, as it bypasses traditional credential theft defenses and exploits the inherent trust in OAuth authentication flows.
Jun 29, 2026
Microsoft warned that **consent phishing** campaigns are increasingly abusing the OAuth authorization flow in Microsoft 365 to trick users into granting attacker-controlled applications access to corporate cloud data. In campaigns observed against remote workers, attackers sent phishing emails through compromised domains or cloud providers, then redirected victims to malicious Office 365 OAuth apps disguised as legitimate services. One campaign impersonated Mexico’s tax administration service to target Spanish-speaking users, while another used financial-themed lures aimed at investment teams; Microsoft later described similar tactics using a **"Review Doc(s) & Sign"** lure tied to a fake business services brand. If a user approved the requested permissions, the attackers could obtain **access and refresh tokens** and use them to access email, files, contacts, SharePoint resources, OneDrive for Business data, and mailbox forwarding settings without directly stealing passwords. Microsoft said it had taken down malicious infrastructure and disabled rogue OAuth apps, while urging organizations to tighten Microsoft identity platform consent settings, limit user consent to **verified publishers** and low-risk permissions, audit application grants, and use Microsoft Defender for Office 365 and related tooling to detect and remediate **illicit consent grants**.
Aug 14, 2026
The FBI warned of an ongoing targeted OAuth consent-phishing campaign in which attackers impersonate government officials, media figures, public personalities, event organizers, and other trusted contacts through phishing emails and commercial messaging applications. The activity, observed since late 2025, targets prominent individuals as well as their relatives and acquaintances, luring them into authorizing attacker-controlled applications. Granting consent gives attackers persistent, high-level access to cloud accounts, including email and sensitive data, through legitimate OAuth authorization flows rather than stolen passwords. The access can survive password changes and bypass conventional password and multifactor authentication protections until the victim explicitly revokes the malicious application's permissions in account security settings. The FBI advised users to verify unsolicited contacts, authorize only trusted applications, and closely review requested OAuth permissions.
Sep 11, 2026