Researchers built a WeChat worm that hijacks accounts while the phone is still ringing
Calif's WeWorm abused a memory corruption bug in WeChat's VoIP stack to take over accounts on an iPhone 17e and two Pixel 10a handsets, and Tencent blocked it server-side on 28 August without publishing an advisory or a CVE.
The Watch · Security desk
What happened
- Calif built WeWorm, a proof-of-concept worm that exploits a memory corruption bug in WeChat's VoIP stack and takes over an account through an incoming call the victim never answers or touches.
- In the demo, a Pixel 10a called an iPhone 17e and seized its WeChat while the handset was still ringing, then the compromised iPhone called a second Pixel 10a and did the same to it.
- Exploitation finishes within seconds and hands the attacker full control of the WeChat account: reading and sending messages, placing calls, and acting as the account's owner.
- Tencent shipped WeChat 8.0.77 for Android and 8.0.76 for iOS on 21 August, and Calif confirmed on 28 August that the exploit was also blocked on Tencent's servers for all users.
- As of 8 September there was no CVE identifier for the flaw and no advisory on Tencent's security response site, and neither company has said which WeChat versions were vulnerable.
Why it matters
- What it limits constraint Without an affected-version list, an operator cannot establish whether a handset was exposed in July or August, and with no published indicators there is no way to test whether a call was ever placed at one. The only assurance available is Tencent's server-side block.
- Who's exposed exposure The WeChat builds for HarmonyOS, Windows, Mac and Linux ship on separate schedules and sit outside what has been confirmed fixed, because neither Calif nor Tencent has said whether the VoIP flaw reached them.
- What's now possible capability The friend-list requirement prices the first victim, not the campaign: each hijacked account becomes an authorised caller for everyone in its contact list, so contact-list membership behaves like a privilege an attacker can acquire.
- What doesn't add up contradiction Calif's AI-speed argument rests on a nine-day figure its own dated log stretches to 19, so how much of the capability-democratisation case you accept depends on which set of numbers the firm defends at disclosure.
Next story · Security · 3h ago
Public exploit takes root on OpenBSD 7.9 through a bug that shipped as a reliability fixConfirmed2 publishers
Memory corruption in the VoIP stack, reached by placing a call [1]. The exploit runs while the handset rings and completes in seconds [7]. Picking up does not interrupt it, and the person who answers hears nothing [2]. Declining ends that attempt only; per Calif, the attacker calls back later, for example while the target is asleep [2].
The one gate is membership of the victim's WeChat friend list [3]. Calif built the demo to show that the gate opens from the inside: "an attacker can compromise one of your friends first and use their account to reach you," the researchers wrote, and the extra trust WeChat extends to contacts then works for the attacker [4]. The chain ran once in the lab, from a Pixel 10a to an iPhone 17e and on to a second Pixel 10a [6].
On its own, the bug takes the account, not the phone [5]. Calif says it can be chained with separate Android or iOS vulnerabilities to reach broader device control [5], and The Hacker News notes that Calif's post describes routes an attacker could use rather than routes it tested [22]. Nothing published shows the worm crossing out of WeChat into the handset. What it does show is a self-propagating account takeover, and for many users that account holds payments, official accounts and mini programs inside the same app [14].
Calif dates its own work: the engineering team knew of the bug on 23 July, finished an Android exploit on 30 July, an iOS exploit on 2 August, and a cross-platform worm demo on 11 August [15][16]. That is 19 calendar days from bug to worm [29]. The public framing is about two days from discovery to first RCE with AI help, plus another week for the worm, or roughly nine [8]; the post does not say whether the shorter figures count working time only [9]. Tencent's client releases landed 29 days after the 23 July date, and the server-side block was confirmed at 36 [27]. Calif shared its complete analysis and working exploits with Tencent on 3 September, after both [23].
Tencent reported 1.439 billion combined monthly active users for Weixin and WeChat as of 30 June 2026 [12]. Calif's report says actors "can compromise over a billion phones (or accounts), upending livelihoods and breaking communities worldwide" [26], and the firm says it published to push governments and technology companies toward cooperation on AI security [13]. Tencent's own contribution to the record is thinner: the iOS release notes and App Store entry describe 8.0.76 as bug fixes, and the most recent post on its security response site is dated April 2022 [18]. No attacks using the flaw have been reported, and Calif does not claim any [20].
Seven days from bug awareness to a working Android RCE against a widely deployed VoIP stack is the figure in Calif's timeline that stands whatever share of the credit AI gets [28].
What to watch
- Whether Tencent publishes affected WeChat versions or requests a CVE, which would let users establish exposure for July and August.
- Calif's promised conference analysis, the first material that would give defenders something searchable.
- Calif says it is examining the same attack surface in other messaging apps and working with their developers; the next disclosure sets the pattern.
Clarity's read
What the record supports and how the coverage leans. The claims behind it follow.
Reality
- Evidence55
- Adoption
- Not scored
- Hype gap+35
- Incentives60
- Confidence60
Perspective Coverage
6 publishers- Builder
- Builder 32%
- Operator
- Operator 51%
- Investor
- Investor 17%
Claim ledger
Ranked by verification strength, evidence, and original report placement.
- [1]
Calif's proof-of-concept worm, dubbed WeWorm, exploits a memory corruption vulnerability in WeChat's Voice-over-IP (VoIP) stack.
- [2]
The victim does not need to answer the call or interact with the phone; if they do answer they hear nothing and the exploit still succeeds; declining stops that attempt but the attacker can try again later, for example while the victim is asleep.
- [3]
The exploit requires the calling account to already be on the victim's WeChat contact/friend list.
- [4]
Calif said the friend-list requirement is not much of a barrier: "an attacker can compromise one of your friends first and use their account to reach you," and once a contact is taken over the extra trust WeChat gives to contacts works for the attacker rather than the user.
- [5]
By itself the demonstrated bug compromises WeChat, not the entire device; Calif says it can be chained with separate Android or iOS vulnerabilities to potentially achieve broader device control.
- [6]
The demo used three phones, two Pixel 10a Android devices and an iPhone 17e: the first Android called the iPhone and took over its WeChat while it was still ringing, and the compromised iPhone then called the second Pixel 10a and compromised it the same way.
- [7]
Calif says exploitation completes within seconds and gives full control of the WeChat account, including reading and sending messages, placing calls and impersonating the victim.
- [8]
Calif said it worked with AI to find the bug and write the first exploit capable of running code on the phone in about two days, and that building the worm took another week.
- [9]
Calif's post does not say whether the shorter (two-day, one-week) figures count only working time.
- [10]
Tencent released WeChat version 8.0.77 for Android and 8.0.76 for iOS on 21 August.
- [11]
Calif confirmed on 28 August that the exploit had also been mitigated server-side for all users, and says the block runs on Tencent's servers so it does not require users to install anything.
- [12]
Tencent put the combined monthly active users of WeChat and Weixin at 1.439 billion as of 30 June 2026 in its second-quarter results.
- [13]
Calif says its goal in publishing is to raise awareness and encourage governments and technology companies to work together on AI security, and argues AI is making advanced attack capabilities available to less-skilled attackers.
- [14]
WeChat's App Store listing covers payments, official accounts and mini programs inside the app.
- [15]
Calif's own timeline: its engineering team knew of the bug on 23 July, the first Android exploit was finished on 30 July, and the worm demo was on 11 August.
- [16]
Calif completed an iOS version of the exploit on 2 August.
- [17]
Checks on 8 September found no CVE identifier for the flaw and no advisory on Tencent's security response site.
- [18]
Tencent has published no advisory about the flaw; its release notes for the iOS version and its App Store entry describe the update as only bug fixes, and its security response site lists its latest announcement as April 2022.
- [19]
Calif is withholding technical details for a future conference presentation; it has published nothing a defender could search for, and there is no way for a user to tell whether they were called.
- [20]
No attacks using the flaw have been reported, and Calif does not say there were any.
- [21]
Calif says it is researching similar weaknesses in other messaging apps and working with developers to reduce those risks, with some changes requiring cooperation from platform owners.
- [22]
The Hacker News notes that Calif's post describes routes an attacker could use rather than ones it tested.
- [23]
Calif shared its complete analysis and working exploits with Tencent on 3 September.
- [24]
Neither Calif nor Tencent has published which WeChat versions were affected, so a user cannot check whether the version they ran in July or August was one of them.
- [25]
Tencent also ships WeChat clients for HarmonyOS, Windows, Mac and Linux on their own release schedules, and neither company has said whether the flaw reached any of them.
- [26]
Calif's report states: "If exploited, actors can compromise over a billion phones (or accounts), upending livelihoods and breaking communities worldwide."
- [27]
Tencent's client releases came 29 days after 23 July, and the server-side block was confirmed 36 days after 23 July.
- [28]
Calif's dates put seven days between its team learning of the bug and a finished Android exploit.
- [29]
From 23 July, when Calif's team knew of the bug, to the 11 August worm demo is 19 calendar days, against roughly nine days implied by Calif's "about two days" plus "another week" framing.
Sources
5 independent publishers whose own reporting we read for this story.
- cyberinsider.comZero-click worm spreads on iPhones and Android via WeChat calls
1 article · September 8, 2026
- helpnetsecurity.com“Zero-click” WeChat worm could hijack accounts and spread via a single call
2 articles · September 8, 2026
- infosecurity-magazine.comResearchers Build WeChat Zero-Click Worm Hijacking Phones via Calls
1 article · September 9, 2026
- scworld.comNew WeWorm tool hacks Android and iOS phones via WeChat calls
1 article · September 9, 2026
- securityaffairs.comWeChat Worm Can Hijack Accounts Without Victims Answering Calls
1 article · September 8, 2026
- thehackernews.comWeChat Zero-Click Worm Took Over Accounts on iPhone and Android via Incoming Calls
2 articles · September 8, 2026
Topics and entities
Topics
- Vulnerability Disclosure
- Zero-click exploits
- AI-assisted vulnerability research
- Consumer messaging app security