Alongside its big public push for AI investments, the U.K. government is also playing a virtual card to catapult itself into the 21st century. Today it announced plans to launch a secure digital wallet to manage government-issued credentials, alongside a chatbot — built in collaboration with OpenAI — to interact with the main GOV.UK portal. Today, it dubbed the chatbot ‘chat-uk’ and said it will be a part of a wider portal and app.
The first two cards it plans to add to its wallet will be a virtual driver’s license, and a virtual veteran card for people who have served in the military. It plans to launch the service later this year, it said in a press conference today. The GOV.UK portal will be online in 2025, but it has already started testing the chat service, and it’s also already opening up for ordinary users to test it, said Peter Kyle, the U.K.’s Secretary of State for technology — even if some of the responses a little odd.
“It is really important the public start to understand that increasingly, in an online world, and government is moving to an online world, that we will interact at an earlier stage so that we can use the power, the insight, and the volume of interactions that the public can provide in testing,” Kyle said in a press conference today in London. “So yes, the chat system did start talking French midway through a testing series, but it was a mistake we learned from.”
The developments come at a crossroads for AI advances.
On one hand, the Labour government has doubled down on the idea of building out an AI economy in the country. Partnerships with private AI companies to invest more in their operations here, more infrastructure such as data centres and a supercomputer to support AI services, and a big commitment from the government to invest in AI services itself. First up, today it unveiled a raft of new AI tools, all still in development: a multi-functional AI assistant for government employees called “Humphrey”; a push to build more consumer-facing AI tools; and more data sharing between government departments to help them build and run AI services.
On the other hand, there remain many questions about how AI services will work in the years ahead. It was only a year ago that the U.K. took a leading role only a year ago in a wider global conversation about AI safety: how the new wave of services being built by companies like OpenAI, Anthropic, Mistral and many others will impact jobs, user privacy and data protection, copyright, and how AI might get misused for nefarious purposes such as to create misinformation, in aid of malicious hacking, and more.
Today, Kyle claimed that in its testing of its new chatbot, there “hasn’t been a single instance” of the chatbot yet being “jailbroken” or returning false information in its “heavy” testing so far.
A push for more digital services in the U.K. comes at the same time as the U.S. is also doubling down on the role tech will be playing in that country. Yesterday, President Trump, on his first day in office, made official his new government “efficiency” effort, dubbed DOGE and led by technology entrepreneur Elon Musk. Trump also repealed an executive order from his predecessor President Joe Biden that sought to reduce AI risks to government, consumers, and businesses. That means that setting up and running safety tests around AI systems are no longer required.
The government said that its new GOV.UK Wallet allows users to securely store government-issued documents on their phone and use them easily when needed, it said. This could be more convenient when, say, you are on the move, and you want to only have a phone and no physical wallet.
The digital driving license, Kyle said today, will also be used to improve online and offline safety. In one example, he described how digital IDs could be used to provide age verification for certain online services, which has been problematic to secure to date. Notably, the U.K. is not exactly ahead of the game here. France has been offering a digital identity app since 2022.
“The technology will make use of security features that are built into modern smartphones, including facial recognition checks similar to those used when people pay using a digital bank card,” it said. “It means that digital documents will be more secure, even if a device is lost.”
There are no plans to make its digital wallet or other digital services compulsory, Kyle added, but he said they are working on the “compelling” and “desirable” nature of the services to get people to use them.
“Why can’t we aspire to have a satisfactory experience where people engage at the moment,” he said.
Keep reading the article on Tech Crunch
Friend, a startup creating a $99, AI-powered necklace designed to be treated as a digital companion, has delayed its first batch of shipments until Q3.
Friend had planned to ship devices to pre-order customers in Q1. But according to co-founder and CEO Avi Schiffman, that’s no longer feasible.
“As much as I would liked to have shipped in Q1 of this year, I still have refinements to do, and unfortunately you can only start manufacturing electronics when you are 95% done with your design,” Schiffman said in an email to customers. “I estimate that by the end of February, when our prototype is complete, that we will begin our final sprint.”
An email I sent out to all Friend preorder customers: pic.twitter.com/wUPR0OhpI4
— Avi (@AviSchiffmann) January 20, 2025
Friend, which has an eight-person engineering staff and $8.5 million in capital from investors including Perplexity CEO Aravind Srinivas, raised eyebrows when it spent $1.8 million on the domain name Friend.com. This fall, as part of what Schiffman called an “experiment,” Friend debuted a web platform on Friend.com that allowed people to talk to random examples of AI characters.
Reception was mixed. TechRadar’s Eric Schwartz noted that Friend’s chatbots often inexplicably kicked off conversations with anecdotes of traumas, including muggings and firings. Indeed, when this reporter visited Friend.com Monday afternoon, a chatbot named Donald shared that the “ghosts of [his] past” were “freaking him the f— out.”
In the above-mentioned email, Schiffman also said that Friend would be winding down its chatbot experience.
“We’re glad that millions got to play around with what I believe to be the most realistic chatbot out there,” Schiffman wrote. “This has really proven our internal ability to manage traffic, and has really taught us a lot about digital companionship … [But] I want us to stay focused on solely the hardware, and I have realized that digital chatbots and embodied companions don’t mix well.”
AI-powered companions have become a hot-button topic. Character.AI, a chatbot platform backed by Google, has been accused in two separate lawsuits of inflicting psychological harm on children. Some experts have expressed concerns that AI companions could worsen isolation by replacing human relationships with artificial ones, and generate harmful content that can trigger mental health conditions.
Keep reading the article on Tech Crunch
On December 21, 2024, just before 2 pm, scientists made the dead speak. ELIZA, the world’s first chatbot is back. Long imitated, but not perfectly replicated, ELIZA has long been thought lost. But scientists discovered an early version of its code in the archives of its creator in 2021 and have spent the intervening years piecing it back together.
ELIZA has been reanimated and you can download it here to see for yourself.
Coded and iterated from 1964 to 1967, ELIZA was developed by MIT computer scientist Joseph Weizenbaum. Rudimentary by today’s standards, ELIZA was a hit at the time of its creation. He gave it the personality of a psychotherapist and his secretary was so enchanted by it, that she asked Weizenbaum to leave the room when she chatted with it.
A new scientific paper from members of the ELIZA Archeology Project details how they found and resurrected the chatbot as well as its origins and subsequent dissemination. Weizenbaum programmed ELIZA in an early language called MAD-SLIP on a time-sharing computer system called the Compatible Time-Sharing System or CTSS.
ELIZA quickly got away from Weizenbaum. As it disseminated through early computer networks, programmers adapted it into other languages. One of these early clones was built in Lisp by one of the technical leads of ARPAnet, the precursor to the modern internet. The Lisp version of Eliza was one of the first bits of data on this nascent network and it spread fast.
“As a result, Cosell’s Lisp ELIZA rapidly became the dominant strain, and Weizenbaum’s MAD-SLIP version, invisible to the ARPAnet, was left to history,” the paper said. “Until it was rediscovered in 2021, the original MAD-SLIP ELIZA had not been seen by anyone for at least 50 years.”
A decade later, a magazine called Creative Computing published an ELIZA clone written in BASIC. It was 1977, the same year that the Apple II, Commodore Pet, and TRS-80 all hit the market. Those machines led to an explosion in home computing and the proliferation of the BASIC computing language.
“And probably not a small number of those hobbyists were interested enough by the possibility of AI to type in this BASIC ELIZA (which was only a couple of pages of code), and experiment with it themselves,” the scientists said. “Because of its brevity and simplicity, and the personal computer explosion, this ELIZA begat hundreds of knock-offs through the decades, in every conceivable programming language, making it perhaps the most knocked-off program in history/ Just as Cosell’s Lisp ELIZA spread via the ARPANet, the BASIC ELIZA spread via the explosive contagion of personal computers.”
There are countless versions of this BASIC version of ELIZA online right now and the original MAD-SLIP version was long thought lost to history. Then Stanford computer scientist Jeff Shrager convinced MIT archivists to root through boxes of Weizenbaum’s material and they made a critical discovery: early versions of the MAD-SLIP code.
The code was incomplete, and it took a lot of tinkering and complicated emulation to get it running again. “It required numerous steps of code cleaning and completion, emulator stack installation and debugging, non-trivial debugging of the found code itself, and even writing some entirely new functions that were not found in the archives or in the available MAD and SLIP implementations,” the paper said.
It took time and a lot of effort, but the code archaeologists got ELIZA working again and they’ve made it available for anyone to play with. “This has been tested on various Linux and MacOS versions, but we’ve noticed some issues with different versions, so your mileage may vary,” they said in the paper. “If you get it working on your machine and find that you have to change something, let us know.”