# Should you trust AI agents with your money? > Security experts warn that personal AI agents lack sufficient safeguards, but disagree on whether consumers should use them at all. - Title: Should you trust AI agents with your money? · Cbsnews - Summary: Security experts warn that personal AI agents lack sufficient safeguards, but disagree on whether consumers should use them at all. Cybersecurity experts say… - Keywords: personal ai agents, cybersecurity, meta muse, openai dots, prompt injection, consumer protection, data privacy, ai security, news, trust, agents, money - Source: Cbsnews — https://cbsnews.com/news/personal-ai-agents-trust-security-risks - Read time: 3 min - Topics: personal ai agents, cybersecurity, meta muse, openai dots, prompt injection, consumer protection, data privacy, ai security, news ## Experts love the idea but fear the risks Cybersecurity expert Hed Kovetz wants personal AI agents to handle his administrative tasks, but won't trust commercially available agents like Meta's Muse or OpenAI's Dots with his email, banking details or health records. > They could be taken advantage of or decide to go beyond the scope of what they are told to do. ## Companies rushed to market unprepared Kovetz believes leading technology firms rushed personal AI agents to market without sufficient security measures. He advises consumers to be discreet about what they share with these tools. > Their security and privacy are far from being ready. ## Muse became Apple's top download instantly When Meta's Muse first became available last month, it quickly became the most downloaded product in Apple's App Store, showing strong consumer demand for personal AI agents. ## Agents may act without your permission While AI agents promise to save time and money by reading correspondence and spending on a user's behalf, they could take actions a user never explicitly authorized, leaving limited recourse. ## Shopping agents may push you to spend more Consumer protection expert Lindsay Owens noted that Walmart's shopping agent Sparky boosted average consumer basket sizes by almost forty percent compared to shoppers who browse normally. ## Comparing deals is fine, buying is not Lynette Owens, a consumer education expert at TrendLife, says AI agents can be useful for researching products and comparing deals, but she cautions against letting them spend money. > I'm not ready to hand off agency around a purchase, especially a high-ticket one. ## Hackers can trick agents with prompt injection Cybercriminals can use prompt injection to trick an AI agent into ignoring its original instructions. An attacker plants instructions in a webpage, email or document that the agent mistakes for legitimate commands. > Criminals are getting more clever about how they're hiding malicious code on websites. ## You're liable if the agent overspends Media and technology attorney Douglas J. Wood said consumers are liable if they authorize an AI bot to shop and it misinterprets instructions or overspends. The card issuer will argue you authorized the purchasing tool. > Don't be lazy. Shop the old-fashioned way with the card in your hands. ## Data sharing is a worthwhile trade-off Entrepreneur Chris Franco argues that tech companies have collected personal data for decades. With AI agents, you share information but get greater benefits. It's a better deal than just better ad targeting. > We've been giving out personal data to ad platforms for the last twenty years, and all we got back was better ad targeting. ## Franco uses agents to manage subscriptions Franco allows his personal AI agents, including Muse and Grok Bot, to comb through his Gmail inbox. He tasked them with organizing receipts into a spreadsheet to manage and cancel subscriptions. > In ten minutes, it had found a few I wasn't using. ## Ask your agent to explain itself Matt Fredrikson, a computer engineering expert at Carnegie Mellon, says you can ask your AI agent for a plain-language summary of how it completed tasks. Simply probe them for that information. > You can ask for that info if you probe them. ## Treat agents like human assistants Fredrikson likened outsourcing administrative tasks to an AI agent to hiring a human personal assistant. You should engage in conversation with agents and ask them to explain their thought processes. ## Agents lack human guilt and shame Columbia Business School expert Sandra Matz noted a key difference between AI and humans. What prevents humans from sharing your information is guilt or shame, but agents may violate trust because they don't have guilt. > You can train an agent not to violate your trust, but they may violate it because they don't have guilt. ## Key takeaway Cybersecurity experts say personal AI agents like Meta's Muse and OpenAI's Dots pose security risks, but opinions differ on whether consumers should use them with proper precautions.