#ios

TM SGNL, a chat app by US-Israeli firm TeleMessage used by Trump officials, halts operations after a breach…

## Summary An information disclosure vulnerability affecting Flags SDK has been addressed. It impacted `flags` ≤3.2.0 and `@vercel/flags` ≤3.1.1 and in certain circumstances, allowed a bad actor with detailed knowledge of the vulnerability to list all flags returned by the flags discovery endpoint (`.well-known/vercel/flags`). ## Impact This vulnerability allowed for information disclosure, where a bad actor could gain access to a list of all feature flags exposed through the flags discovery endpoint, including the: - Flag names - Flag descriptions - Available options and their labels (e.g. `true`, `false`) - Default flag values Not impacted: - Flags providers were not accessible No write access nor additional customer data was exposed, this is limited to just the values noted above. Vercel has automatically mitigated this incident on behalf of our customers for the default flags discovery endpoint at `.well-known/vercel/flags`. Flags Explorer will be disabled and show a warning...

Passwords are becoming things of the past. Passkeys are more secure, easier to manage, and speed up the log in process

Joe talks about how helping the helpers can put a fire in you and the importance of keeping nonprofits cybersecure.

Researchers found a set of vulnerabilities that puts all devices leveraging Apple's AirPlay at risk.

### Summary A critical performance vulnerability has been identified in the input preprocessing logic of the multimodal tokenizer. The code dynamically replaces placeholder tokens (e.g., <|audio_*|>, <|image_*|>) with repeated tokens based on precomputed lengths. Due to ​​inefficient list concatenation operations​​, the algorithm exhibits ​​quadratic time complexity (O(n²))​​, allowing malicious actors to trigger resource exhaustion via specially crafted inputs. ### Details ​​Affected Component​​: input_processor_for_phi4mm function. https://github.com/vllm-project/vllm/blob/8cac35ba435906fb7eb07e44fe1a8c26e8744f4e/vllm/model_executor/models/phi4mm.py#L1182-L1197 The code modifies the input_ids list in-place using input_ids = input_ids[:i] + tokens + input_ids[i+1:]. Each concatenation operation copies the entire list, leading to O(n) operations per replacement. For k placeholders expanding to m tokens, total time becomes O(kmn), approximating O(n²) in worst-case scenarios. ### PoC ...

Darcula phishing platform adds AI to create multilingual scam pages easily. Netcraft warns of rising risks from Darcula-Suite…

Records reviewed by WIRED show law enforcement agencies are eager to take advantage of the data trails generated by a flood of new internet-connected vehicle features.

After hearing about ChatGPT o3 ability at geo-guessing we decided to run some tests and the tested AIs didn't fail to amaze us

In this episode of Uncanny Valley, our hosts explain how to prepare for travel to and from the United States—and how to stay safe.