Encrypted instructions can fool AI assistants like Grok and Gemini

<p class="wp-block-paragraph">A new type of prompt injection attack shows why giving AI assistants access to browsers, code tools, and private data deserves extra caution.</p>

<p class="wp-block-paragraph">AI researchers <a href="https://adversa.ai/blog/cryptographic-context-injection-grok-data-theft/" rel="noreferrer noopener nofollow" target="_blank">describe</a> “Cryptographic Context Inject