First autonomous AI agent is here, but is it worth the risks?

2 weeks ago 16

If you haven’t heard the buzz astir Manus yet, it's the caller AI exemplary unveiled by a Singapore-based institution called Butterfly Effect. This isn’t conscionable different chatbot. It’s 1 of the archetypal genuinely autonomous AI agents, capable to bash its ain research, marque decisions and adjacent transportation retired plans, each with hardly immoderate quality oversight.

Sounds impressive, right? But here’s the thing: While each this innovation opens up breathtaking possibilities, it besides brings immoderate superior privateness and information questions. Whether you’re anxious to effort retired the latest AI oregon you’d alternatively steer clear, it’s worthy knowing what Manus could mean for your idiosyncratic information and integer safety.

STAY PROTECTED & INFORMED! GET SECURITY ALERTS & EXPERT TECH TIPS – SIGN UP FOR KURT’S ‘THE CYBERGUY REPORT’ NOW

First autonomous AI cause is here, but is it worthy the risks?

First genuinely autonomous AI cause (Manus) (Kurt "CyberGuy" Knutsson)

Your information could beryllium utilized to bid AI, adjacent if you spell obscurity adjacent it

This has already been happening for a portion with different AI models. There person been several class-action lawsuits against tech companies for scraping idiosyncratic information for use successful AI training, with 1 suit claiming that OpenAI uses "stolen backstage information, including personally identifiable information, from hundreds of millions of net users, including children of each ages, without their informed cognition oregon consent."

Many companies besides merchantability oregon stock information with 3rd parties, including for the purposes of AI training. For example, X has been utilizing idiosyncratic information to bid its AI, Grok, and is besides sharing idiosyncratic information with 3rd parties to bid AI models. So whether you usage AI (including Manus AI erstwhile it’s released to the public) oregon not, there’s inactive a bully accidental that they person your idiosyncratic information.

First genuinely autonomous AI cause (Manus) (Kurt "CyberGuy" Knutsson)

GOOGLE'S NEW AI TECH MAY KNOW WHEN YOUR HOUSE WILL BURN DOWN

Why should you attraction if AI has your data?

It boils down to information privateness and security. When AI companies cod idiosyncratic accusation – things similar your name, email address, location location, societal media posts and adjacent photos – they often store it successful ample databases to bid their models. If those databases aren’t decently secured, they tin go golden mines for hackers. Think of them arsenic integer "honey pots": saccharine and tempting to cybercriminals due to the fact that of each the idiosyncratic information inside.

This isn’t conscionable a theoretical risk. Just this year, different AI company, DeepSeek, experienced a large leak that exposed much than a cardinal delicate records. When this benignant of accusation gets out, it tin extremity up connected the acheronian web wherever scammers and individuality thieves are much than blessed to scoop it up. That’s however radical extremity up dealing with everything from individuality theft and impersonation to tons of spam calls, phishing emails and targeted scams.

But information leaks aren’t the lone danger. Even erstwhile the accusation is secure, it tin beryllium misused oregon manipulated successful ways that are hard to detect. One of the biggest caller threats is thing called a punctual injection attack.

In elemental terms, punctual injection is simply a mode for attackers to instrumentality an AI into doing thing it wasn’t expected to. Researchers precocious recovered that attackers tin embed hidden commands successful seemingly harmless text, similar a website bio, email signature oregon adjacent a societal media post. When the AI reads that substance during grooming oregon browsing, it tin accidentally travel the hidden instructions, which mightiness archer it to leak information, alteration however it behaves oregon springiness unauthorized entree to backstage data.

First genuinely autonomous AI cause (Manus) (Kurt "CyberGuy" Knutsson)

DANGERS OF OVER-SHARING WITH AI TOOLS

Manus AI poses adjacent bigger risks

Unlike astir AI tools that respond to commands you springiness them directly, Manus is autonomous, meaning it tin enactment connected its own, travel links, work contented and marque decisions without quality approval. While that sounds beauteous cool, it’s besides what makes it a batch harder to spot erstwhile thing goes wrong.

If Manus encounters malicious content, it could unknowingly travel a hidden prompt, compromise your information oregon change its behavior, and determination whitethorn beryllium nary evident motion that thing has happened until it’s excessively late. There are besides immoderate concerns astir wherever Manus sends and stores data. Security researchers person traced information backmost to servers successful Shenzhen, China, raising reddish flags astir surveillance, jurisdiction and who mightiness yet person entree to your information.

So adjacent if Manus isn’t scraping thing excessively delicate today, the operation of anemic oversight, almighty automation and questionable information practices makes it acold riskier than your mean AI assistant.

We reached retired to Manus AI for a remark connected this nonfiction but did not perceive backmost earlier our deadline.

First genuinely autonomous AI cause (Manus) (Kurt "CyberGuy" Knutsson)

GENAI, THE FUTURE OF FRAUD AND WHY YOU MAY BE AN EASY TARGET

So, however tin you support yourself?

As I’ve mentioned, whether you usage Manus AI oregon not, your idiosyncratic accusation volition inactive apt extremity up successful its datasets, truthful the champion mode to support yourself is to marque your accusation hard to find. Here are immoderate things you tin do.

1) Be cautious astir utilizing Manus: With that level of autonomy, particularly since it’s a precise caller AI (not to notation the inconsistency successful its privateness argumentation regarding governance), I wouldn’t urge utilizing it rather yet. There’s conscionable excessively overmuch harm it tin bash if it malfunctions oregon is taxable to a punctual injection attack.

2) Keep your idiosyncratic accusation nether wraps: With truthful galore companies selling and trading idiosyncratic information, galore of them to tech companies to bid AI, it’s a bully thought to withhold your information unless you perfectly can’t debar sharing it.

3) Actively region your idiosyncratic accusation from the net truthful your information can’t beryllium scraped by AI models similar Manus: Set your societal media to private, bounds information postulation done privateness settings and region your accusation from information brokers and people-search sites. Autonomous cloud-based agents specified arsenic Manus are designed to extract and analyse immense amounts of online data, and your publically disposable accusation tin beryllium swept into their grooming datasets without your cognition oregon consent.

While nary work promises to region each your information from the internet, having a removal work is large if you privation to perpetually show and automate the process of removing your accusation from hundreds of sites continuously implicit a longer play of time. Check retired my apical picks for information removal services here.

4) Use beardown antivirus bundle to assistance support against the broader onslaught aboveground created by autonomous AI systems: Manus’ quality to enactment independently and interact with outer systems increases the hazard of adversarial attacks and malware transmission, truthful robust endpoint information is indispensable to support your devices and data.

The champion mode to safeguard yourself from malicious links that instal malware, perchance accessing your backstage information, is to person antivirus bundle installed connected each your devices. This extortion tin besides alert you to phishing emails and ransomware scams, keeping your idiosyncratic accusation and integer assets safe. Get my picks for the champion 2025 antivirus extortion winners for your Windows, Mac, Android and iOS devices.

5) Use privacy-focused products and services: Whenever you cannot debar sharing idiosyncratic information, opt for companies that prioritize your privacy. For example, usage the DuckDuckGo browser alternatively of Chrome and Brave oregon Proton Mail alternatively of Gmail. Check retired the champion backstage and harmless browser alternatives to Google.

6) Keep each bundle updated, including your operating system, browsers and information tools, to spot vulnerabilities that attackers mightiness exploit: As AI agents similar Manus run successful the unreality and whitethorn interact with your systems, unpatched bundle tin go an casual people for exploitation oregon unauthorized access. Therefore, it's ever a bully thought to support software, operating systems, browsers and information tools updated astatine each times.

7) Enable multifactor authentication (MFA) connected each your accounts: As AI-driven attacks and automated information extraction go much common, MFA adds a important furniture of defense, making it overmuch harder for attackers to compromise your accounts, adjacent if an AI strategy leaks oregon scrapes your credentials.

8) Use strong, unsocial passwords for each account: As cloud-based AI agents similar Manus tin interact with outer systems and perchance entree delicate data, unsocial passwords assistance forestall a breach of 1 relationship from starring to a broader compromise. Consider utilizing a password manager to make and store analyzable passwords. Get much details astir my best expert-reviewed password managers of 2025 here.

Kurt's cardinal takeaways

As we determination guardant with AI similar Manus, it’s breathtaking to deliberation astir the possibilities, but it's besides important to support our eyes unfastened astir what it means for our idiosyncratic data. Even if you ne'er usage Manus yourself, your info could inactive extremity up successful its grooming data.

So what tin you do? Stay informed, instrumentality complaint of your integer footprint and take tools that prioritize your privacy. As AI keeps evolving, being proactive astir your information is the champion mode to bask the benefits of innovation portion keeping your idiosyncratic info safe.

Do you deliberation companies down almighty AI agents similar Manus should beryllium required to travel stricter privateness rules and autarkic oversight earlier they’re allowed to cod and usage people’s data? Let america cognize by penning america at Cyberguy.com/Contact.

For much of my tech tips and information alerts, subscribe to my escaped CyberGuy Report Newsletter by heading to Cyberguy.com/Newsletter.

Ask Kurt a question oregon fto america cognize what stories you'd similar america to cover .

Follow Kurt connected his societal channels:

Answers to the most-asked CyberGuy questions:

New from Kurt:

Kurt "CyberGuy" Knutsson is an award-winning tech writer who has a heavy emotion of technology, cogwheel and gadgets that marque beingness amended with his contributions for Fox News & FOX Business opening mornings connected "FOX & Friends." Got a tech question? Get Kurt’s escaped CyberGuy Newsletter, stock your voice, a communicative thought oregon remark astatine CyberGuy.com.

Read Entire Article