OpenAI has issued a warning about the risks of prompt injection attacks in AI browsers, which can manipulate AI agents into executing harmful commands. To combat these threats, OpenAI is employing a large language model (LLM)-based adversary to test and identify potential vulnerabilities. Users are advised to restrict agent access as a precautionary measure. In addition to OpenAI, companies like Anthropic and Google emphasize the importance of implementing layered security measures to safeguard against such attacks. This development is particularly relevant for traders considering investments in AI-related crypto assets, as they should carefully evaluate the risk-to-reward ratio in light of these security concerns.