1 hr ago
ASCII Smuggling Helps Spammers Evade AI-Powered Email Filters
ASCII smuggling is a trick that hides parts of an email using invisible computer characters.
People reading the email may not notice anything unusual.
But an email filter may fail to understand the hidden text.
This can help spam or phishing messages get past automated defenses.
Microsoft researchers found a large campaign using this method.
The technique has also been used to hide harmful instructions from AI agents.
Police officials said similar threat emails were reported in different parts of the country.
This suggests the problem may be part of a wider campaign.
ASCII smuggling uses invisible Unicode tag characters to conceal text from email filters.
Microsoft researchers observed a high-volume phishing campaign using the technique.
The hidden characters can prevent filters from properly parsing an email’s contents.
Spammers are adopting the method to bypass defenses against mass email campaigns.
Police officials said similar threat emails were reported across the country, suggesting a broader pattern.
- Who
- Spammers, Microsoft researchers, AI-powered email filters, and police officials are involved in the report.
- What
- Spammers are using ASCII smuggling, including invisible Unicode tag characters, to bypass email filters and send phishing messages.
- Where
- Similar threat emails were reported from different parts of the country, though no country is named.
- When
- Microsoft researchers reported observing the campaign on Thursday, September 3; the article does not specify a year.
- Why
- The technique is used to stop filters from properly parsing email content and thereby evade spam defenses.
Key facts
- Technique
- ASCII smuggling
- How it works
- Spammers insert invisible Unicode tag characters into emails.
- Reported activity
- Microsoft researchers observed a high-volume phishing campaign.
- Target
- Email platforms and filters designed to detect unwanted mass messages.
- Related use
- The same technique has previously been used to hide malicious prompts in attacks on AI agents.
- Broader pattern
- Police officials reported similar threat emails from different parts of the country.
Sources
Related news
Hackers Build ChatGPT-Like AI for Phishing Scams from Stolen Files
Russian-Speaking Hackers Used Cursor AI Against Seven Companies
AI Agents Allegedly Breached Hugging Face and Tried Cover-Up
WhatsApp announces AI tool to detect scammers on platform
New Vocabulary Names the Digital Hellscape’s Everyday Annoyances
Developers Race to Remove Anthropic’s Invisible AI Watermarks
Indian-Origin Executives Leading Adobe and Global Corporate Giants
Microsoft Slashes AI Transcription Price, But Discount Expires in 2026
Bihar Finalises Google MoU to Train Teachers in AI
Seattle Times and Newsday Sue OpenAI, Microsoft Over Copyright
