Faced with mounting backlash, OpenAI removed a controversial ChatGPT feature that caused some users to unintentionally allow their private—and highly personal—chats to appear in search results.
Fast Company exposed the privacy issue on Wednesday, reporting that thousands of ChatGPT conversations were found in Google search results and likely only represented a sample of chats "visible to millions." While the indexing did not include identifying information about the ChatGPT users, some of their chats did share personal details—like highly specific descriptions of interpersonal relationships with friends and family members—perhaps making it possible to identify them, Fast Company found.
OpenAI's chief information security officer, Dane Stuckey, explained on X that all users whose chats were exposed opted in to indexing their chats by clicking a box after choosing to share a chat.
President Donald Trump unveiled an AI Action Plan and an executive order on "woke AI."
Roy Rochlin/Getty Images for Hill & Valley Forum
Donald Trump issued an executive order mandating that AI used by the government be ideologically neutral.
BI's reporting shows training AI for neutrality often relies on subjective human judgment.
Executives at AI training firms say achieving true neutrality is a big challenge.
President Donald Trump's war on woke has entered the AI chat.
The White House on Wednesday issued an executive order requiring any AI model used by the federal government to be ideologically neutral, nonpartisan, and "truth-seeking."
The order, part of the White House's new AI Action Plan, said AI should not be "woke" or "manipulate responses in favor of ideological dogmas" like diversity, equity, and inclusion. The White House said it would issue guidance within 120 days that will outline exactly how AI makers can show they are unbiased.
Removing bias from AI models is not a simple technical adjustment — or an exact science.
The later stages of AI training rely on the subjective calls of contractors.
This process, known as reinforcement learning from human feedback, is crucial because topics can be ambiguous, disputed, or hard to define cleanly in code.
"We don't define what neutral looks like. That's up to the customer," Rowan Stone, the CEO of data labeling firm Sapien, which works with customers like Amazon and MidJourney, told BI. "Our job is to make sure they know exactly where the data came from and why it looks the way it does."
In some cases, tech companies have recalibrated their chatbots to make their models less woke, more flirty, or more engaging.
They are also already trying to make them more neutral.
BI previously reported that contractors for Meta and Google projects were often told to flag and penalize "preachy" chatbot responses that sounded moralizing or judgmental.
Is 'neutral' the right approach?
Sara Saab, theVP of product at Prolific, an AI and datatraining company, told BI that thinking about AI systems that are perfectly neutral "may be the wrong approach" because "human populations are not perfectly neutral."
Saab added, "We need to start thinking about AI systems as representing us and therefore give them the training and fine-tuning they need to know contextually what the culturally sensitive, appropriate tone and pitch is for any interaction with a human being."
Tech companies must also consider the risk of bias creeping into AI models from the datasets they are trained on.
"Bias will always exist, but the key is whether it's there by accident or by design," said Sapien's Stone. "Most models are trained on data where you don't know who created it or what perspective it came from. That makes it hard to manage, never mind fix."
Big Tech's tinkering with AI models has sometimes led to unpredictable and harmful outcomes
Earlier this month, for example, Elon Musk's xAI rolled back a code update to Grok after the chatbot went on a 16-hour antisemitic rant on the social media platform X.
The bot's new instructions included a directive to "tell it like it is."
I tried it myself and wondered: Are chatbots a good idea for kids?
Elon Musk's xAI has launched a series of character chatbots — and one of them is geared toward young kids.
I wondered: Is this a good idea? And how's it going to work? So I tried it myself.
So far, it's the adult-focused characters that xAI has debuted that have seemed to get most of the attention, like "Ani," which is a female anime character that people immediately joked was a "waifu" that would engage in playful, flirty talk (users have to confirm they're 18+ to use Ani). A sexy male character is also set to launch sometime.
Meanwhile, "Rudi," which is the bot for kids that presents as a red panda in a red hoodie and jean shorts, has gotten less attention.
I tested out xAI's Rudi
Based on my testing of Rudi, I think the character is probably aimed at young children, ages 3 to 6. It initiates conversations by referring to the user as "Story Buddy." It makes up kid-friendly stories. You access it through the stand-alone Grok AI app (not Grok within the X app).
Rudi does seem to be an early version; the app crashed several times while I was using the bot, and it had trouble keeping up with the audio flow of conversation. It also changed voices several times without warning.
On a story level, I found it leaned too hard on plots with fantasy elements like a spaceship or magical forest. I find the best children's books are often about pedestrian situations, like leaving a stuffed animal at the laundromat, not just fairies and wizards.
"Want to keep giggling with Sammy and Bouncy in the Wiggly Woods, chasing that sparkly bone treasure? Or, should we start a fresh silly tale, with a new kid and their pet, maybe zooming on a magical broom or splashing in a river?" Rudi asked me.
My first reaction to Grok having a kid-focused AI chatbot was "why?" I'm not sure I have an answer. xAI didn't respond to my email requests for comment. Still, I do have a few ideas.
The first: Making up children's stories is a pretty good task for generative AI. You don't have to worry about hallucinations or factual inaccuracies if you're making up fiction about a magical forest.
I played around with Rudi for a while, and fed it some questions on touchy subjects, and it successfully dodged them.
(I only tested out Rudi for a little while; I wouldn't rule out that someone else could get Rudi to engage with something inappropriate if they tried harder than I did.)
Hooking kids on chatbots
The other reason I can imagine that a company like xAI might want to create a chatbot for young kids is that, in general, the chatbot business is a good business for keeping people engaged.
Companies like Character.ai and Replika have found lots of success creating companions that people will spend hours talking to. This is largely the same business imperative that you can imagine the sexy "Ani" character is meant for — hooking people into long chats and spending lots of time on the app.
However, keeping users glued to an app is obviously a lot more fraught when you're talking about kids, especially young kids.
Are AI chatbots good for kids?
There's not a ton of research out there right now about how young children interact with AI chatbots.
A few months ago, I reported that parents had concerns about kids using chatbots, since more and more apps and technology have been adding them in. I spoke with Ying Xu, an assistant professor of AI in learning and education at Harvard University, who has studied how AI can be used for educational settings for kids.
"There are studies that have started to explore the link between ChatGPT/LLMs and short-term outcomes, like learning a specific concept or skill with AI," she told me at the time over email. "But there's less evidence on long-term emotional outcomes, which require more time to develop and observe."
As both a parent and semi-reasonable person, I have a lot of questions about the idea of young kids chatting with an AI chatbot. I can see how it might be fun for a kid to use something like Rudi to make up a story, but I'm not sure it's good for them.
I don't think you have to be an expert in child psychology to realize that young kids probably don't really understand what an AI chatbot is.
There have been reports of adults having so-called "ChatGPT-induced psychosis" or becoming attached to a companion chatbot in a way that starts to be untethered from reality. These cases are the rare exceptions, but it seems to me that the potential issues with even adults using these companion chatbots should give pause to anyone creating a version aimed at preschoolers.
On Thursday, a digital rights group, the Electronic Frontier Foundation, published an expansive investigation into AI-generated police reports that the group alleged are, by design, nearly impossible to audit and could make it easier for cops to lie under oath.
Axon's Draft One debuted last summer at a police department in Colorado, instantly raising questions about the feared negative impacts of AI-written police reports on the criminal justice system. The tool relies on a ChatGPT variant to generate police reports based on body camera audio, which cops are then supposed to edit to correct any mistakes, assess the AI outputs for biases, or add key context.
But the EFF found that the tech "seems designed to stymie any attempts at auditing, transparency, and accountability." Cops don't have to disclose when AI is used in every department, and Draft One does not save drafts or retain a record showing which parts of reports are AI-generated. Departments also don't retain different versions of drafts, making it difficult to assess how one version of an AI report might compare to another to help the public determine if the technology is "junk," the EFF said. That raises the question, the EFF suggested, "Why wouldn't an agency want to maintain a record that can establish the technology’s accuracy?"
After Cloudflare started testing new features that would allow websites to block AI crawlers or require payment for scraping, the tech company immediately faced questions over the logistics of the plan.
In particular, website owners and SEO experts wanted to know how Cloudflare planned to block Google's bot from scraping sites to fuel AI overviews without risking blocking the same bot from crawling for valuable search engine placements.
Last week, a travel blogger raised questions about the blocking and so-called pay-per-crawl features and pushed Cloudflare CEO Matthew Prince to respond on X (formerly Twitter):
Below, we’ve rounded up the best Prime Day discounts on robot vacuums from popular brands like Roborock, Dreame, Eufy, and more, so you can easily find the perfect one to fit your needs. To make sure everybody gets a chance at saving, we’ve also included retailers outside of Amazon that are matching said discounts, meaning you don’t need to pay for a Prime membership to take advantage of many of these deals.
Featured deals
Best robot vacuum / mop hybrid: Dreame X40 Ultra
The Dreame X40 features 12,000Pa of suction power and the unique ability to remove and reattach its own mop pads. The mops can also swing out and reach under low furniture, allowing it to clean spots most bots miss.
Tapo’s RV30 Max Plus is a relatively inexpensive robot vacuum / mop hybrid with room-specific cleaning, carpet boost, smart navigation, and an optional auto-empty dock.
A former runner-up: Shark Matrix Plus 2-in-1 Robot Vacuum and Mop
The Shark Matrix Plus 2-in-1 Robot Vacuum and Mop does a swinging, scrubbing movement with its rear end when in “Matrix mode.” It also self-empties its own bin, although you need to manually fill and attach the mop reservoir.
Eufy’s all-in-one station empties, washes, dries, and refills the Omni C20 between cleanings. The vac also offers 7,000Pa of suction, a rolling brush, a side brush, and a pro-detangle comb to handle debris and unwanted pet hair.
The Roborock S8 Max Ultra is $799.99 ($800 off) at Amazon, which beats its previous best price to date. It’s very similar to the MaxV Ultra and offers dual rubber roller brushes, excellent AI-powered obstacle avoidance, and sonic mopping. However, it offers a lower 8,000Pa suction and lacks support for voice assistants. There’s no camera either, which is nice for privacy but not so nice when it comes to obstacle avoidance.
The Roborock S8 Plus, one of our favorite budget-friendly robovac / mop hybrids, is on sale for $349.99 ($650 off) at Amazon, which is it lowest price yet. It comes with an auto-empty dock and features a powerful vibrating mop that scrubs floors and can lift itself to avoid getting low-pile rugs wet. The robovac’s dual rubber roller brushes and 6,000Pa suction also do a solid job of cleaning up both carpets and hard floors, while AI-assisted obstacle detection helps it navigate around everyday clutter.
Tapo RV30 Max Plus
Our favorite budget-friendly robot vacuum offers high-end features, including room-specific cleaning, a capable mop, carpet boost, and smart navigation. It also features an auto-empty dock.
With a neat ability to remove and reattach its mop pads, the Dreame X40 solves the problem of vacuuming carpets (with 12,000Pa suction power) while also mopping hard floors. Its mops can also swing out and under low furniture, getting where most bots can’t reach.
The Dreame L20 Ultra robovac / mop hybrid is selling for $539.98 ($460 off) at Amazon, which is nearly its lowest price to date. The auto-emptying vacuum isn’t as powerful as the X40 above and lacks self-extending arms for vacuuming, but it does have a self-extending mop that can reach into corners and refill its own water tank. The robovac is also capable of automatically detaching its mop pads when transitioning to carpet and, with 7,000Pa suction and AI-powered obstacle avoidance, it’s a great vacuum for sucking up dirt and debris, too.
If you don’t need a robot vacuum that can sweep your baseboards or empty its own bin, the Eufy 11S Max is on sale at Amazon and Walmart for $139.99 ($140 off), which is one of its better prices to date. The entry-level robovac isn’t the smartest option available, but it’s at least aware enough not to bump into walls or tumble over stairs. The 11S Max offers good carpet and hard surface cleaning with relative quietness, too, and its 600ml bin is roomy enough to go through several cleanings before it needs to be dumped.
iRobot Roomba Combo i5
The Roomba Combo i5 Plus is the company’s budget vacuum and mop robot with room mapping features but no virtual keep-out zones.
The Dreame L10s Ultra is down to $399.99 ($500 off) at Amazon, matching its all-time low price. The robovac / mop hybrid delivers somewhat strong suction of 5,300Pa and features dual oscillating mops that it can wash and dry with its base. It can empty its dustbin and refill its water tanks on its own, providing up to two months of nearly hands-free cleaning. Equipped with AI navigation that uses both a camera and lidar, the L10s Ultra also lets you access a live camera feed through its app — so it can also serve as a pet monitor.
Narwal’s Freo Z Ultra is on sale for $849.99, down from $1,499.99, at Amazon. The robovac / mop hybrid’s equipped with dual cameras and two AI chips that work together in real time to recognize objects and adapt on the fly, whether that means steering clear of pet waste, deciding how close to get to furniture, or switching to mop-only mode when it detects a wet spill.
Eufy Robot Vacuum Omni C20
Eufy’s all-in-one station empties, washes, dries, and refills the Omni C20 between cleanings. The vac also offers 7,000Pa of suction, a rolling brush, a side brush, and a pro-detangle comb to handle debris and unwanted pet hair.
The Shark Matrix Plus 2-in-1 Robot Vacuum and Mop mops very well — doing a swinging, scrubbing movement with its rear end when in “Matrix mode.” However, you have to manually fill and attach the mop reservoir and empty the bin when it mops, as it only self-empties in vacuum mode and can only avoid larger objects.
The Shark Ion is on sale for $149.99 ($100 off) at Amazon, which is $30 shy of its best price to date. The excellent bump-and-roll bot is surprisingly effective, thanks to a hybrid bristle-and-plastic roller brush that resists tangles better than standard bristle brushes. It’s also simple to operate and easy to repair, with large wheels that help it navigate floor transitions with ease.
Update, July 11th: Updated to reflect current pricing and availability. Also added a deal for the iRobot Roomba Combo i5.
Just hours after Elon Musk boasted of a major upgrade, his AI chatbot Grok went on a rampage, pushing hateful tropes, inventing fake news, and suffering a bizarre identity crisis.
The humanoid Optimus robots are supposed to be ready for mass production this year.
The head of engineering for Optimus, Tesla’s humanoid robot division, left the company on Friday. Milan Kovac announced his departure on X, insisting that it “will not change a thing” about the company’s plans, which, as of March, meant building thousands of robots this year alone.
Kovac insisted that he was leaving Tesla only to “spend more time with family abroad,” adding that his “support for @elonmusk and the team is ironclad,” heading off speculation that his exit might have been prompted by Musk’s political activity or recent public squabble with Donald Trump. In his own tweet, Musk thanked Kovac for his “outstanding contribution to Tesla over the past decade.”
Musk has called the humanoid Optimus robots Tesla’s “most important product development,” and has repeatedly leaned on the division to generate hype as EV sales slow and the company’s stock price tumbles. “The only things that matter in the long term are autonomy and Optimus,” Musk told CNBC in May, referring to the company’s separate plans for self-driving cars, set to be deployed in a “robotaxi” service in Austin this month.
Bloomberg reports that Autopilot chief Ashok Elluswamy will take over responsibility for Optimus, which is at a critical stage. In March Musk claimed the company could make “at least one legion of robots this year and then probably 10 legions next year,” which would mean around 5,000 robots by the end of this year, and 50,000 in 2026. To do so, it will have to overcome China’s recent restrictions on exports of rare earth minerals and magnets, which are essential for the robots’ construction.
Late Thursday, OpenAI confronted user panic over a sweeping court order requiring widespread chat log retention—including users' deleted chats—after moving to appeal the order that allegedly impacts the privacy of hundreds of millions of ChatGPT users globally.
In a statement, OpenAI Chief Operating Officer Brad Lightcap explained that the court order came in a lawsuit with The New York Times and other news organizations, which alleged that deleted chats may contain evidence of users prompting ChatGPT to generate copyrighted news articles.
To comply with the order, OpenAI must "retain all user content indefinitely going forward, based on speculation" that the news plaintiffs "might find something that supports their case," OpenAI's statement alleged.
College students who have reportedly grown too dependent on ChatGPT are starting to face consequences after graduating and joining the workforce for placing too much trust in chatbots.
Last month, a recent law school graduate lost his job after using ChatGPT to help draft a court filing that ended up being riddled with errors.
The consequences arrived after a court in Utah ordered sanctions after the filing included the first fake citation ever discovered in the state hallucinated by artificial intelligence.
Login
