WhatsApp closes loopholes for hijackers: long password instead of PIN and multiple passkeys at once
The messenger is moving two-factor verification to full-fledged passwords with letters and symbols, allows linking more than one passkey to an account, and is increasingly warning about suspicious calls and attempts to link someone else's device. At the same time, Meta is testing an AI detector for scam messages that analyzes conversations right on the phone.

Why WhatsApp is tackling security again
Account takeover in a messenger is one of the most unpleasant types of fraud: a person loses not just access to their chats, but effectively their digital identity, and their contacts receive messages asking for money on their behalf. WhatsApp decided not to limit itself to yet another warning and reworked several protection mechanisms at once — from the way sign-in is confirmed to the analysis of incoming calls and messages.
The overall logic behind the changes is simple: instead of shifting all the responsibility onto the user's attentiveness, the platform itself starts to hint that a situation looks suspicious.

Two-step verification: a full-fledged password instead of a short PIN
Currently, the additional sign-in protection is built on a six-digit PIN code. That's better than nothing, but such a code has a weak spot: it's short, easy to brute-force, and it barely helps if a fraudster has already tricked a person into giving up the SMS confirmation code and is trying to entrench themselves in the account.
After the update, users will be able to replace the PIN with a longer password — with letters, numbers, and special characters. The point is to create a second barrier in case the first one has already been breached: even after gaining access to the sign-in code, an attacker will hit an additional line of defense that can't be cracked in minutes.
What this changes in practice
- Hacking through brute force becomes noticeably more time-consuming — and therefore less worthwhile.
- The social engineering tactic of "read me the code from the SMS" stops working on its own: the password is also needed, and the fraudster doesn't have it.
- There's now an incentive to finally set up a password manager — few people are willing to keep a long string of characters in their head.
Passkeys: now you can link several
Another area is the shift to passwordless sign-in methods. Users on both Android and iOS will be able to add not just one but several passkeys to their account. The technology itself relies on what's already in the device: face recognition, a fingerprint scanner, or screen lock.
According to the platform, more than a million people worldwide already use this sign-in method. The logic behind multiple keys is simple: one is linked to the phone, the second to a tablet or backup device. If the gadget is lost, broken, or sent in for repair, access isn't wiped out along with it.

Android will tell you exactly who's calling
A separate set of changes concerns calls. On Android, the messenger will start showing more context about a caller from an unknown number: for example, whether the person is calling from another country or whether they share a group with you.
This is a small thing that closes a very common scenario: a fraudster poses as customer support or a "work acquaintance," and the victim can't quickly check whether they have even any point of contact in common. Shared groups and geography are rough but useful signals.
AI against scam messages
Meta is testing a feature in the messenger called Scam Alert, built on artificial intelligence. It can scan messages from unknown contacts and look for characteristic signs of deception.
An important detail: the check happens right on the user's device, not on servers. If the system sees something suspicious, it doesn't just issue a warning but also suggests specific actions: block the chat or send a report.
Earlier, the app already introduced warnings about suspicious device-linking requests — exactly the scenario where a fraudster tries to connect someone else's account to their own gadget.

A shift in emphasis — and what you should do
The main thing worth noting: the fight against fraud is moving from the level of "the user must be attentive" to the level of "the platform itself warns and checks." This is a sensible turn. People get tired, get distracted, read messages on the go — and that's exactly when they fall for it.
But no automatic filter replaces basic habits. The most reliable things remain:
- never forward sign-in and confirmation codes to anyone, no matter who they supposedly "belong" to according to the story;
- be wary if a stranger rushes you, plays on your emotions, or asks you to follow a link;
- enable two-step verification and, when the option becomes available, choose a long password instead of a six-digit PIN;
- add a second passkey on a backup device so you don't depend on a single gadget.
Technical improvements close some of the loopholes, but the final step — agreeing to a suspicious action — is still made by you. That's exactly why it's worth spending ten minutes on your settings now rather than dealing with a hijacked account later.
Frequently asked questions
Related materials
All materials
Validating Candidates Costs More Than Generating Them: Honeypots Against LLM Vulnerability Hunting
The study examines the cost asymmetry in agentic vulnerability discovery: generating candidate versions is cheap, while confirming them through reachability analysis, execution, and PoC development consumes most of the budget, making verification a potential standalone target for defense. The proposed RedHerring—repository-adapted decoys with a hidden certificate—reduced the number of real vulnerabilities found by 38.7–60.4% at equal cost across 33 OSS-Fuzz projects; even after the agent was warned, the reduction remained at 37.2%.
9 November 2026
Checking LLM Hallucinations via SQL-Based Source Representation Without Fine-Tuning
The study examines whether an LLM can apply a basic symbolic skill—building a SQL database from source documents—to ground its outputs in sources without supervision when detecting hallucinations. On RAGTruth and DiaHalu, this neuro-symbolic pipeline proves more accurate than direct prediction and comparable to state-of-the-art methods, without requiring domain-specific fine-tuning.
9 November 2026
MathAdv: What the Final Accuracy of Machine-Generated Proofs Conceals
A diagnostic suite of tasks spanning 13 undergraduate and graduate mathematics topics evaluates machine-proving systems in Lean 4 not only by their final results, but also by their subject-matter mastery, the quality of their informal reasoning, and their robustness to rephrased conditions. A component-by-component analysis shows that translating a problem into formal form remains the main obstacle, and that success depends heavily on the area of mathematics and how the problem is framed.
9 November 2026