Personalized AI chatbots go off the rails on Instagram and Messenger
Users can now talk to an artificial intelligence (AI) version of a psychologist, a cooking expert, a version of Batgirl, a parody of Vladimir Putin or even a "virtual girlfriend." All of these bots are created by the platforms' users, based on Llama, Meta's large language model. It only takes a few seconds to set up a chatbot by describing its purpose and personality, with the option to add further instructions if needed. The user can then decide whether to make their chatbot publicly accessible or restrict it to their friends only.
However, Meta has warned that "we review AIs you create" before they are published, to ensure they do not violate the platform's rules. Yet the review process is rather superficial. Naruto, Harry Potter, the French YouTuber Squeezie, God, Marilyn Monroe, Breaking Bad 's Walter White and even France's prime minister, François Bayrou: Le Monde quickly found many chatbots that impersonate real people, fictional characters covered by copyright protections, and religious figures – including several Jesus Christ bots – all of which are forbidden by Meta. While some of these bots appeared not to have interacted with any users, others have already exchanged millions of messages.
Furthermore, despite rules prohibiting the chatbots from posing as financial advisers, dozens of AI bots dedicated to cryptocurrency, personal finance or investment could be found easily.
The few profiles tested by Le Monde only repeated general statements. However, in just a few moments, we were able to create a bot specialized in financial advice that encouraged beginners to invest online via the fraudulent trading website RiveGarde, which a previous Le Monde investigation found was connected to an organized crime network. It was then possible to share the chatbot with friends after it went through Meta's automatic verification process, which took less than a minute.
Hashtags
Try Our AI Features
Explore what Daily8 AI can do for you:
Comments
No comments yet...
Related Articles
LeMonde
4 hours ago
- LeMonde
Capturing CO₂ through cryogenics: Eiffage pioneers new technology to decarbonize its industry
The installation – a metal tower 11 meters high, flanked by a container and a tank, one for storing liquid nitrogen and the other for capturing carbon dioxide (CO 2) – would almost go unnoticed beside the two enormous lime kilns that continuously transform limestone, extracted from a quarry owned by Bocahut, into calcium oxide. This network of pipes and tubes, perpetually covered in white dust, is the demonstrator for the Lyon-based start-up Revcoo, which has developed a proprietary process for capturing CO 2. Since 2024, it has been tested on this 120-hectare open-air site, owned by French construction group Eiffage and located in Haut-Lieu (northern France), 25 kilometers south of Maubeuge. "For now, our pilot has a capture capacity of 1,000 metric tons of CO 2 per year. Our goal is to multiply that by 10 by 2027, then reach 80,000 to 100,000 metric tons in 2030 – the equivalent of the site's total emissions," explained Hugo Lucas, founder and president of Revcoo. The 33-year-old engineer founded the start-up in 2019 with the ambition of contributing to the decarbonization of heavy industry. His patented "CarbonCloud" technology is relatively straightforward: It captures smoke off the factory's chimneys, sorts out the CO 2 by freezing it, liquefies it, and then stores it in tanks. The entire process runs on electricity, with no water or solvents required.
Euronews
13 hours ago
- Euronews
Meta removes 6.8 million WhatsApp accounts linked to criminal scammers
WhatsApp has taken down 6.8 million accounts that were 'linked to criminal scam' centres targeting people online around that world, its parent company Meta said. The account deletions, which Meta said took place over the first six months of the year, arrive as part of wider company efforts to crack down on scams. In a Tuesday announcement, Meta said it was also rolling new tools on WhatsApp to help people spot scams, including a new safety overview that the platform will show when someone who is not in a user's contacts adds them to a group, as well as ongoing test alerts to pause before responding. Scams are becoming all too common and increasingly sophisticated in today's digital world. Too-good-to-be-true offers and unsolicited messages attempt to steal consumers' information or money, with scams filling our phones, social media, and other corners of the internet each day. Meta noted that 'some of the most prolific' sources of scams are criminal scam centres, which often span from forced labour operated by organised crime – and warned that such efforts often target people on many platforms at once, in attempts to evade detection. That means that a scam campaign may start with messages over text or a dating app, for example, and then move to social media and payment platforms, Meta said. Meta, which also owns Facebook and Instagram, pointed to recent scam efforts that it said attempted to use its own apps – as well as TikTok, Telegram, and AI-generated messages made using ChatGPT – to offer payments for fake likes, to enlist people into a pyramid scheme, or to lure others into cryptocurrency investments. Meta linked these scams to a criminal scam center in Cambodia and said it disrupted the campaign in partnership with ChatGPT maker OpenAI.
Euronews
16 hours ago
- Euronews
A UK politician turned himself into an AI chatbot. Here's how it works
Imagine being able to corral your local politician and grill them about their views, anytime you fancy – no appointments, no red tape. That's exactly what one British Member of Parliament has set out to achieve. Mark Sewards, a Labour Party politician who represents Leeds South West and Morley, has unveiled an artificial intelligence (AI) clone of himself that is designed to answer questions and assist with constituents' queries around the clock. 'I'm very keen to explore how new technologies, such as AI, can help strengthen the connection between an MP's office and the constituents we serve,' Sewards said in a statement. He encouraged his constituents to 'give AI Mark a try' – but noted in an Instagram post that the tool isn't designed to replace his official duties. Rather, the idea is to provide a 24/7 service where constituents can raise local issues, ask about policy matters, or drop a message to their MP's office without waiting weeks for a formal reply. The bot 'simply gives people another option to contact me, any time of day,' Sewards wrote. How a tech firm cloned Sewards The Leeds-based startup Neural Voice developed Seward's AI twin, dubbing him the 'UK's first virtual MP'. The company says it uses cutting-edge neural voice synthesis and advanced natural language processing to help create voice AI assistants for phone systems and websites. It used voice recordings of Sewards in parliamentary sessions, social media profiles, and previous correspondences with constituents to create a digital double that mirrors his speaking style, tone, and mannerisms with uncanny accuracy, the company said. It hopes the new technology will make local politicians more accessible to their constituents. "For too long, politicians have been out of reach for their constituents,' said Jeremy Smith, Neural Voice's CEO and co-founder. 'The vast majority of people do not know the name of their MP, let alone their voice and what they stand for,' Smith added. He believes this technology could soon become par for the course, with other MPs and even businesses potentially following suit in a bid to bridge the communication gap. Digital doubles in politics and business AI is increasingly being deployed in public-facing roles to overcome physical barriers and create constant feedback loops. Last year, the mayor of a Japanese city, Yokosuka, created an AI avatar of himself to improve communication with the thousands of English-speaking US Naval personnel based there. Meanwhile, a political outfit in Denmark called the Synthetic Party created an AI candidate called 'Leader Lars' to stand in the country's 2022 elections, though the party did not win votes. The AI model, which the party says was fed with material from fringe political parties in Denmark dating back to 1970, was designed to engage with citizens who typically do not vote and to demonstrate AI's role in democracy. Companies in Poland and China have also experimented with having AI CEOs front their brands. Yet such tools can be repurposed to impersonate public figures without their approval, or to generate convincing fake personas that appear to be real public figures. High-profile deepfake scams have targeted Italian defense minister Guido Crosetto, US Secretary of State Marco Rubio, and several celebrities, including Taylor Swift and Joe Rogan, whose voices were used to promote a scam that promised people government funds. Deepfakes were created every five minutes in 2024, according to a report from think tank Entrust Cybersecurity Institute. Whether a digital doppelgänger can truly replace an in-person chat with a public figure remains to be seen, but back in the United Kingdom, Sewards believes it will be at least one promising option. 'The AI revolution is happening and we must embrace it or be left behind,' he said on the social media platform X.
