Workers are spilling secrets to chabots
The big picture: The problem isn't new, but as workplace genAI use increases, many employers still lack or don't enforce AI policies, causing employees to use bots in secret or without proper training.
By the numbers: Harmonic Security sampled a million prompts and 20,000 files submitted to 300 genAI tools and AI-enabled SaaS applications between April and June.
43,700 of the prompts (4.4%) and 4,400 of the uploaded files (22%) contained sensitive information.
Between the lines: Personal and free chatbot accounts make up a large share of corporate data exposure.
Nearly half (47.42%) of sensitive uploads to Perplexity were from users with standard (non-enterprise) accounts.
About a quarter of the prompts with sensitive information came through the free version of ChatGPT, and another 15% of sensitive prompts were submitted via free versions of Google Gemini accounts.
Zoom in: Overall, including free and paid tiers, ChatGPT was by far the biggest source of prompt-based information exposure, followed by Microsoft Copilot and Google Gemini.
Code was the most common type of sensitive data sent to chatbots.
Harmonic says code was "especially prevalant in ChatGPT, Claude, DeepSeek and Baidu Chat."
The number of prompts containing proprietary code was disproportionately high in Claude, which is often regarded as the best AI tool for coders.
Sensitive prompts to ChatGPT involved M&A planning, financial modeling, and investor communications.
The intrigue: Tools that feel safe — like document editors or design platforms — may now include genAI features trained on user data, creating exposure risk that bypasses traditional controls, Harmonic says.
Harmonic found that Canva, Replit, Grammarly, and other tools with LLMs embedded inside them were used for legal strategy, internal emails, client data, and code.
These uses were often not flagged as AI tools by corporate systems.
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
Bloomberg
30 minutes ago
- Bloomberg
The Challenge Towns Face in Powering AI and Data Centers
As AI's rapid expansion drives a growing need for data centers, small towns like Warrenton, Virginia are pushing back - worried about the noise, energy demands, and the challenges of dealing with big tech companies. Despite the promise of much-needed revenue, Warrenton's residents overwhelmingly opposed an Amazon data center, citing concerns over community impact and energy infrastructure strain. With power needs rising faster than the grid can support, industry leaders say new solutions like on-site clean energy will be essential if data centers are to become welcome neighbors rather than disruptive intrusions. (Source: Bloomberg)
Miami Herald
an hour ago
- Miami Herald
JPMorgan revamps strong forecast on Federal Reserve rate cuts
Buzzy rumblings are escalating that the Federal Reserve is on the cusp of a dovish pivot to lower interest rates in your wallet and household budget. JPMorgan has revised its outlook as to when the Federal Reserve will cut interest rates. Don't miss the move: Subscribe to TheStreet's free daily newsletter JPMorgan cited mounting uncertainty around President Donald Trump's temporary Fed appointee, plus growing signs of weakness in the labor market. Economists and market watchers are concerned over what is perceived as the president's efforts to politicize the independent central bank's monetary policymaking. Related: Trump makes surprise decision on Federal Reserve There's also uncertainty around the impact of President Trump's tariffs on inflation. In addition, many worry the Trump's administration's aggressive immigration campaign is decimating the low-wage and blue-collar work force, while the AI revolution wipes out entry-level roles for many new college graduates. "This is an unusual administration to say the least,'' Kevin O'Leary of "Shark Tank" told CNN. Image source: Watson/Getty Images President Trump announced on Aug. 7 that he was nominating Council of Economic Advisers Chair Stephen Miran to temporarily fill a vacant Federal Reserve Board seat. Miran, a Trump loyalist, has joined the president and Treasury Secretary Scott Bessent in heavily criticizing the Fed and Chair Jerome Powell over interest rates. Miran, if confirmed by the Senate this fall, will serve in an interim appointment lasting until Jan. 31, 2026. The Harvard-educated labor economist advocates stronger control over the Fed by the executive branch, shorter terms for the Board of Governors and nationalizing regional Federal Reserve banks. Analysts warned that his proximity to the president's aggressive stance on interest rates may increase a growing internal divide within the Federal Open Market Committee, the Fed's policymaking panel that sets benchmark interest rates. Related: A divided Federal Reserve mulls interest rate cut after wild week Danske Bank's chief strategist Frederik Romedahl told Bloomberg that Miran's appointment "adds some uncertainty," even if it's not a total game-changer. Market watchers like Andrew Brenner have deemed Miran controversial and questioned his business experience. Others say the Miran appointment is unlikely to affect the expected pace of rate cuts in the short term because of the macroeconomic dynamics rippling through the economy. Long term, the president could gain two appointees to the Fed board in 2025. First, he needs a permanent replacement for Miran's role (which very well could be Miran, depending on the president's whims). Second, Powell's term as chair ends in May, and he has said he hasn't decided if he plans to serve out the rest of his term into 2028. Historically, Fed chairs leave the board when their chair terms expire. Theoretically, that would give President Trump four seats on the seven-member board. Fed Governors Michelle Bowman and Christopher Waller are both Trump appointees from his first administration. Waller is considered the current top candidate to replace Powell next year. The Federal Reserve's dual congressional mandate requires monetary policy that balances low unemployment and low inflation using interest rates as the benchmark tool. Higher interest rates lower inflation but increase job losses. Lower interest rates decrease unemployment but increase inflation. More Federal Reserve: GOP plan to remove Fed Chair Powell escalatesTrump deflects reports on firing Fed Chair Powell 'soon'Former Federal Reserve official sends bold message on 'regime change' The president's personal and professional attacks on Powell have escalated since the FOMC voted in July to hold the benchmark Federal Funds Rate steady at 4.25% to 4.50%. The vote was noteworthy because Waller and Bowman both dissented, saying that emerging weakness in the labor market justified a .25 percentage-point rate cut. That was the first dissenting FOMC vote since 1993. The last rate cut was in December 2024. President Trump wants a 3-point cut, saying it is necessary to curb recession fears, ease the stagnant housing market with lower mortgages, and reduce the interest on the federal deficit. Economists and markets must consider not only data but also volatile political dynamics now swirling around the Fed. JPMorgan had originally forecast one .25 percentage-point rate cut in December 2025. It pulled that forward, forecasting a .25 cut at the Sept. 17 Fed meeting, Reuters reported. JPMorgan added three additional .25 cuts before the Fed pauses. The note did not give a firm timeline for those cuts. Analyst Michael Feroli wrote that "the risk-management considerations at the next meeting may go beyond balancing employment and inflation risks." Meanwhile, market odds for a September cut have surged, with the CME Group's FedWatch tool forecasting the likelihood at 88.9%. Related: Consumers fear inflation as tariffs hit home: Federal Reserve report The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.
Gizmodo
an hour ago
- Gizmodo
The World Will Enter a 15-Year AI Dystopia in 2027, Former Google Exec Says
The world is hurtling towards an inevitable AI dystopia in the very near future, according to Mo Gawdat, the former chief business officer of Alphabet's moonshot factory, formerly known as Google X. 'We will have to prepare for a world that is very unfamiliar,' Gawdat said in an interview on the 'Diary of a CEO' podcast, adding that humanity's key values like freedom, human connection, accountability, reality, and power are all facing a major disruption by AI. And this dystopia isn't far off, we have already started seeing signs of it as of last year and will continue to see an escalation of signs next year, Gawdat said. The beginning of the descent into Gawdat's dystopia, he predicts, will begin in 2027 and last for the next 12 to 15 years. The former Google executive wasn't always of this opinion: the speed with which artificial intelligence technologies have been developing caused him to change his mind and convinced him that this short-term dystopia is inevitable. 'It is completely within our hands to change that, but I have to say, I don't think humanity has the awareness at this time to focus on this,' Gawdat said. But Gawdat says AI is not necessarily the main driver of this dystopia, and especially not in the way most people imagine (that is, existential risks from scenarios that have AI assuming full control). Rather, Gawdat says that AI acts as a magnifier of existing societal issues and 'our stupidities as humans.' 'There is absolutely nothing wrong with AI,' Gawdat said. 'There is a lot wrong with the value set of humanity at the age of the rise of the machines.' Artificial intelligence was not developed to usher in a dystopia, in fact it had a rather utopic mission. By further automating mundane tasks, AI has the potential to ease the workload of millions of workers worldwide in every job and field, potentially giving them back their precious time of the day without sacrificing overall productivity. However, that is not exactly how things are panning out for workers. In a world governed by one value above all others –capitalism– that utopian dream is being warped by the relentless pursuit of profit. AI's disruption of the labor market has already begun, according to some experts, as the technology starts to completely reshape the way we view work. Instead of helping out people's workload, companies that are maximizing productivity with artificial intelligence are laying people off or slowing down hiring to further maximize profit, or asking even more of existing workers. That is no coincidence, according to Gawdat, who believes that all technology ever created magnifies existing human abilities and values, and the biggest value set of humanity currently is capitalism. This disconnect between intended consequences and the reality of the negative downsides has echoed in other technological advancements as well. 'How often did social media connect us and how often did it make us more lonely? How often did mobile phones make us work less? That was the promise, the early ads of Nokia, where people had parties, is that your experience of mobile phones?' Gawdat said. Elon Musk Turns His AI Chatbot Into a Male Fantasy Engine Another thing that AI is going to escalate beyond control, according to Gawdat, is 'the evil that man can do.' For those following along with the news the past year or so, this is no surprise. From AI-generated deepfake porns and AI's increased entrance into warfare to maximize lethality with autonomous weapons and generative AI in the military, the technology has served as an aide to the worst that humanity has to offer. This was on full display this week when Elon Musk's chatbot Grok unveiled a new image-and-video-generation feature, the primary use of which so far has been generating women in heavily sexualized male fantasies. AI-powered scams – and specifically AI-powered crypto scams which is something OpenAI CEO Sam Altman himself had warned against– have skyrocketed. A report from blockchain intelligence firm TRM Labs found that crypto scams were up 456% over the last year thanks to AI deepfake technology. And nuclear war experts are worried that AI could soon power nuclear weapons. Sam Altman Is Right: AI-Powered Crypto Scams Are Exploding AI is also fine tuning public surveillance methods at a massive scale. In a world where there is 'a massive concentration of power,' as Gawdat puts it, that's a major concern. AI-powered public surveillance systems are actively in place in many countries right now, the prime example being the mass surveillance infrastructure in China. It's not just a foreign concern though, the United States government is also now utilizing AI to monitor the social media accounts of immigrants and travelers wishing to enter the country. Despite all of this, AI does continue to usher in remarkable changes for good. Artificial intelligence has already had a measurable impact on scientific discovery and advancements, most notably in medicine and pharmaceutical research. Gawdat does believe that a utopian use of AI is possible down the road thanks to advancements like these. But first, humanity must contend with its pitfalls. 'The bigger picture is to put pressure on governments to understand that there is a limit to which people will stay silent,' Gawdat said, adding that governments should regulate the use of AI rather than AI itself. 'You cannot regulate the design of a hammer so that it can drive nails but not kill anyone, but you can criminalize the killing of a human by a hammer,' Gawdat said. The hammer of AI is now in our hands, and it's here to stay. The only question left is whether we have the will to write the laws against the murder.
