Technologies
Gen AI Chatbots Are Starting to Remember You. Should You Let Them?
An AI model’s long memory can offer a better experience — or a worse one. Good thing you can turn it off.
Until recently, generative AI chatbots didn’t have the best memories: You tell it something and, when you come back later, you start again with a blank slate. Not anymore.
OpenAI started testing a stronger memory in ChatGPT last year and rolled out improvements this month. Grok, the flagship tool of Elon Musk’s xAI, also just got a better memory.
It took significant improvements in math and technology to get here but the real-world benefits seem pretty simple: You can get more consistent and personalized results without having to repeat yourself.
“If it’s able to incorporate every chat I’ve had before, it does not need me to provide all that information the next time,” said Shashank Srivastava, assistant professor of computer science at the University of North Carolina at Chapel Hill.
Those longer memories can help with solving some frustrations with chatbots but they also pose some new challenges. As with when you talk to a person, what you said yesterday might influence your interactions today.
Here’s a look at how the bots came to have better memories and what it means for you.
Improving an AI model’s memory
For starters, it isn’t quite a “memory.” Mostly, these tools work by incorporating past conversations alongside your latest query. “In effect, it’s as simple as if you just took all your past conversations and combined them into one large prompt,” said Aditya Grover, assistant professor of computer science at UCLA.
Those large prompts are now possible because the latest AI models have significantly larger “context windows” than their predecessors. The context window is, essentially, how much text a model can consider at once, measured in tokens. A token might be a word or part of a word (OpenAI offers one token as three-quarters of a word as a rule of thumb).
Early large language models had context windows of 4,000 or 8,000 tokens — a few thousand words. A few years ago, if you asked ChatGPT something, it could consider roughly as much text as is in this recent CNET cover story on smart thermostats. Google’s Gemini 2.0 Flash now has a context window of a million tokens. That’s a bit longer than Leo Tolstoy’s epic novel War and Peace. Those improvements are driven by some technical advances in how LLMs work, creating faster ways to generate connections between words, Srivastava said.
Other techniques can also boost a model’s memory and ability to answer a question. One is retrieval-augmented generation, in which the model can run a search or otherwise pull up documents as needed to answer a question, without always keeping all of that information in the context window. Instead of having a massive amount of information available at all times, it just needs to know how to find the right resource, like a researcher perusing a library’s card catalog.
Read more: AI Essentials: 27 Ways to Make Gen AI Work for You, According to Our Experts
Why context matters for a chatbot
The more an LLM knows about you from its past interactions with you, the better suited to your needs its answers will be. That’s the goal of having a chatbot that can remember your old conversations.
For example, if you ask an LLM with no memory of you what the weather is, it’ll probably follow up first by asking where you are. One that can remember past conversations, however, might know that you often ask it for advice about restaurants or other things in San Francisco, for example, and assume that’s your location. “It’s more user-friendly if the system knows more about you,” Grover said.
A chatbot with a longer memory can provide you with more specific answers. If you ask it to suggest a gift for a family member’s birthday and tell it some details about that family member, it won’t need as much context when you ask again next year. “That would mean smoother conversations because you don’t need to repeat yourself,” Srivatsava said.
A long memory, however, can have its downsides.
You can (and maybe should) tell AI to forget
Having a chatbot recommend a gift poses a conundrum that’s all too common in human memories: You told your aunt you liked airplanes when you were 12 years old, and decades later you still get airplane-themed gifts from her. An LLM that remembers things about you could bias itself too much toward something you told it before.
“There’s definitely that possibility that you can lose your control and that this personalization could haunt you,” Srivastava said. “Instead of getting an unbiased, fresh perspective, its judgment might always be colored by previous interactions.”
LLMs typically allow you to tell them to forget certain things or to exclude some conversations from their memory.
You may also deal with things you don’t want an AI model to remember. If you have private or sensitive information you’re communicating with an LLM (and you should think twice about doing so at all), you probably want to turn off the memory function for those interactions.
Read the guidance on the tool you’re using to be sure you know what it’s remembering, how to turn it on and off and how to delete items from its memory.
Grover said this is an area where gen AI developers should be transparent and offer clear commands in the user interface. “I think they need to be providing more controls that are visible to the user, when to turn it on, when to turn it off,” he said. “Give a sense of urgency for the user base so they don’t get locked into defaults that are hard to find.”
How to turn off gen AI memory features
Here’s how to manage memory features in some common gen AI tools.
ChatGPT
OpenAI has a couple types of memory in its models. One is called “reference saved memories” and it stores details that you specifically ask ChatGPT to save, like your name or dietary preferences. Another, “reference chat history,” remembers information from past conversations (but not everything).
To turn off either of these features, you can go to Settings and Personalization and toggle the items off.
You can ask ChatGPT what it remembers about you and ask it to forget something it has remembered. To completely delete this information, you can delete the saved memories in Settings and the chat where you saved that information.
Gemini
Google’s Gemini model can remember things you’ve discussed or summarize past conversations.
To modify or delete these memories, or to turn off the feature entirely, you can go into your Gemini Apps Activity menu.
Grok
Elon Musk’s xAI announced memory features in Grok this month and they’re turned on by default.
You can turn them off under Settings and Data Controls. The specific setting is different between Grok.com, where it’s “Personalize Grok with your conversation history,” and on the Android and iOS apps, where it’s “Personalize with memories.”
Technologies
Trump says MAGA Inc. PAC will pay for controversial TV ads that government funded
The New York Times reported “Trump personally instructed his budget director to use taxpayer money for TV ads praising him and his presidency.”
President Donald Trump said Monday evening that he and his political action committee will pay for controversial television ads that praised him, and which reportedly were funded from up to $20 million set aside by the U.S. Department of Homeland Security.
The White House later clarified that the super PAC — MAGA Inc. — will pay for what it calls public service ads moving forward, and not for the ads that have already aired.
Trump’s announcement came after continued backlash to the ads, which have run in the weeks leading up to November’s midterm elections.
Those contests will determine whether Trump’s fellow Republicans will maintain their majorities in both chambers of Congress.
Critics say the ads mirror Republican campaign talking points. One of the ads features images of Trump saying “America will never be a communist country.”
“The Radical Left is upset with the fact that I am taking Ads, which I consider to be a positive promotion for our Great U.S.A., and paying for them with U.S.A. money,” Trump said in a post on Truth Social on Monday.
“This is a rather standard thing to do but, rather than doing that, although nothing will make them happy, I have decided to do the Patriotic Ads, among others, and pay for them myself, and with money I raised for MAGA, Inc.,” Trump said.
AdImpact has tracked roughly $9.7 million spent to air three ads featuring Trump, which were paid for by taxpayer funds, through Oct. 5.
Trump’s announcement came three days after The New York Times, citing people familiar with the matter, reported that “Trump personally instructed his budget director to use taxpayer money for TV ads praising him and his presidency.”
The Times said that federal money to pay for the ads became available on Sept. 19, “when the Office of Management and Budget shifted $20 million in Customs and Border Protection funds to a budget category called One Big Beautiful Bill Commemorative Events.” Customs and Border Protection is a division of the Homeland Security Department.
Sen. Maggie Hassan, D-N.H., in a Sept. 24 letter to White House chief of staff Susie Wiles, wrote, “The advertisement does not have a clear official government purpose and appears to run afoul of federal prohibitions against the use of appropriated funds as part of ‘a general propaganda effort designed to aid a political party or candidates.’”
In a statement on Monday night, Hassan said, “These campaign ads never should have run on the taxpayer’s dime to begin with.”
“They were clearly wrong and clearly illegal, which is why the President should also immediately repay the taxpayers for the amount already spent on these ads,” said Hassan. “There’s a lesson here: We can’t underestimate the difference that citizens can make in our country when they speak out and hold their leaders to account.”
Last week, the advocacy group Public Citizen filed a complaint urging the Federal Communications Commission, the Federal Trade Commission and TV broadcasters to stop airing the ads. Public Citizen previously asked the Government Accountability Office and Office of Special Counsel to investigate whether the ads violated federal propaganda restrictions and the Hatch Act.
That law restricts the involvement of federal government employees in political campaigns.
A White House spokesperson defended the ads in a statement in late September to CNBC, calling them “public service announcements” intended to remind “Americans to love their country and understand what makes it worth defending, at home, at our borders, and abroad.”
“The ad is educational and unapologetically patriotic. We should be proud of our country,” the spokesperson said.
MAGA Inc. has raised $424.4 million and spent $32.4 million during the 2025-26 cycle through Aug. 31, leaving the Trump Super PAC with $415.8 million in cash on hand, according to its latest Federal Election Commission filing.
MAGA Inc. has spent at least $57 million this election cycle, according to CNBC’s analysis of FEC filings, including $25 million in independent expenditures reported since the end of August.
— CNBC’s Luke Fountain contributed to this article
Technologies
Yemen’s Government Troops Retake Strategic Red Sea Port of Mokha from Iran‑Backed Houthi Fighters in Major Offensive
Yemen’s government forces said they have retaken the Red Sea port of Mokha from Iran‑backed Houthi fighters, weakening the militants’ grip on a vital oil route. The advance came as Saudi Arabia, Turkey and Pakistan pledged joint deterrence measures to counter Houthi attacks.
Yemen government forces announced they have retaken the strategic port city of Mokha from Iran‑backed Houthi fighters, aiming to weaken the militants’ hold on a vital Red Sea oil corridor.
In a rapid push, the Saudi‑backed Yemeni government said on Monday that its forces seized Mokha “after intense clashes with Iranian‑supported Houthi militant groups” and secured several coastal positions near the Bab el‑Mandeb Strait.
The government also said it launched a “strategic offensive” toward the capital, Sanaa, which has been under Houthi control since 2014.
Verum could not independently verify the claims. The Houthis have reportedly denied that Mokha has fallen.
Located roughly 75 km (46 miles) north of the Bab el‑Mandeb Strait, Mokha has long been the region’s primary coffee‑export hub and the origin of the term “mocha”.
Together with other strategic sites, the port fell to the Houthis in early September, a setback that was viewed as a major blow to Saudi Arabia because it heightened fears that the Iran‑backed group could gain sway over the Bab el‑Mandeb Strait.
Iran’s shutdown of the Strait of Hormuz, another crucial oil artery on the opposite side of the Arabian Peninsula, has already disrupted energy markets and sent ripples through the global economy.
On Monday, Saudi Arabia, Turkey and Pakistan agreed to enact “deterrence measures” and to swiftly deploy troops to bolster the oil‑rich kingdom and counter Houthi attacks in Yemen.
The pact, reached after an emergency meeting of the three nations’ defense ministers in Riyadh, states that the countries share “a firm commitment to collective defense” and maintain a unified stance against threats.
Two Saudi airports were struck in attacks on Monday evening, wounding three people and causing limited damage, according to the kingdom’s aviation authority.
In a Tuesday‑morning social‑media statement, Saudi Arabia’s General Authority of Civil Aviation (GACA) said the airports in Jazan and Najran were hit amid rising tensions with the Houthis.
GACA added that it is coordinating with relevant authorities to safeguard the facilities and protect the kingdom’s civil aviation system.
Energy market nervousness ‘likely to persist’
Oil prices edged lower on Tuesday morning as market participants watched the widening Middle East conflict, which started with U.S. and Israeli strikes on Iran in late February.
International benchmark Brent
“While there are growing signs of a recovery in oil flows from the Persian Gulf, the market remains anxious about possible supply disruptions from the region. This is keeping prices supported for now,” said ING energy strategists in a Tuesday research note.
“Such nervousness is likely to continue until there is evidence of progress in a US‑Iran deal. Meanwhile, the risk of further escalation remains very real,” they added.
Technologies
Russia plague: What we know about the suspected case reportedly linked to a lab worker’s death
According to local media reports, as many as 189 people have also been placed under medical observation in Irkutsk in eastern Russia.
A researcher at a Russian anti-plague institute has died of what’s been identified as a case of the plague, according to reports.
Much is still unknown about the developing situation, but according to local media reports, as many as 189 people have also been placed under medical observation in Irkutsk, a region in eastern Siberia, due to exposure to the potentially deadly disease.
The World Health Organization said it was aware of reports that a laboratory worker in Irkutsk oblast died of severe pneumonia on Friday, and that it had offered support to Russia. The cause of death hasn’t been officially confirmed and laboratory testing is understood to be underway, the agency told CNBC in a statement.
“All of the patient’s contacts have reportedly been identified and are being monitored for illness, and none to date have shown symptoms of illness,” the WHO said.
What is the plague and how does it spread?
Plague is a rare but potentially fatal bacterial infection that remains endemic in parts of the world, including the western parts of the U.S., but can be treated with antibiotics if identified quickly. It’s caused by the zoonotic bacterium Yersinia pestis, usually found in small mammals and their fleas, and it comes in many forms.
Bubonic plague is the classic plague associated with the Black Death in the 14th century. Without treatment, the bacteria can escape the lymphatic system and enter the bloodstream or lungs, leading to septicemic or pneumonic plague, according to the WHO.
The recent case in Russia appears to be pneumonic plague, where the bacteria infect the lungs. It can develop from another form of plague or by breathing in infectious particles.
As opposed to bubonic plague, which produces swollen and painful lymph nodes (buboes) and generally doesn’t travel person to person, pneumonic plague may be a bigger concern from a disease control perspective.
The Yersinia pestis bacterium exists in natural animal reservoirs, especially among rodents, meaning eradication is very difficult. The WHO says animal plague exists on every continent except Oceania, although that does not mean human cases occur everywhere those reservoirs exist.
“Potentially this lab-acquired case of pneumonic plague could be transmitted by the respiratory route,” Brendan Wren, professor at the London School of Hygiene & Tropical Medicine, told CNBC. “Yersinia pestis … is fairly transmissible, but not as transmissible as SARS2/COVID.”
What’s happening with the suspected case in Russia?
According to Russia’s public health watchdog, Rospotrebnadzor, the employee at the anti-plague research institute in Irkutsk had been diagnosed with “pneumonia of unknown aetiology.” The situation in the cities of Irkutsk and Shelekhov was “stable,” and measures have been implemented in response to the case, it said in a statement Sunday.
Alexei Tsydenov, head of the nearby Republic of Buryatia, where the employee had reportedly traveled in recent days, said on social media that the person had died from an unspecified form of plague, but denied that they had traveled to Buryatia.
CNBC has not been able to independently verify the reports. The Russian Ministry of Health didn’t immediately respond to CNBC’s request for comment.
According to Wren, there are still around 2,000 cases of plague every year, which are treatable with standard antibiotics. “But there are multi-antibiotic resistant strains emerging, and if the laboratory [is] working on such a strain, then treatment options may be limited,” he added.
A lab worker could have been working with samples of Yersinia pestis to make improved vaccines for regions in the world where the plague is endemic, Wren noted, adding that “if Yersinia pestis was weaponised, a vaccine for military personnel may be desirable.”
Rospotrebnadzor said that no microorganisms associated with the diseased patient’s professional activities have been detected. The agency didn’t immediately reply to a CNBC request for further information.
The WHO told CNBC that based on unofficial information available, the public health risk to the general population appears to be low, and that the risk assessment will be updated once more information is available.
— CNBC’s Jenny Lee contributed to this report.
-
Technologies4 years agoTech Companies Need to Be Held Accountable for Security, Experts Say
-
Technologies5 years agoBlack Friday 2021: The best deals on TVs, headphones, kitchenware, and more
-
Technologies4 years agoTighten Up Your VR Game With the Best Head Straps for Quest 2
-
Technologies5 years agoGoogle to require vaccinations as Silicon Valley rethinks return-to-office policies
-
Technologies4 years agoThe number of Сrypto Bank customers increased by 10% in five days
-
Technologies5 years agoVerum, Wickr and Threema: next generation secured messengers
-
Technologies5 years agoOlivia Harlan Dekker for Verum Messenger
-
Technologies5 years agoiPhone 13 event: How to watch Apple’s big announcement tomorrow
