WASHINGTON--The Biden administration is confronting the rapidly expanding use of artificial intelligence, warning of the dangers the technology poses to public safety, privacy and democracy while having limited authority to regulate it. Vice President Kamala Harris, who met Thursday with the chief executives of leading AI companies Google, Microsoft, OpenAI and Anthropic, said the technology "has the potential to dramatically increase threats to safety and security, infringe civil rights and privacy, and erode public trust and faith in democracy."

The UK government has announced an initial impact review in response to the continued growth and concerns around generative AI and learning language models. The investigation will reportedly look at how the creation and distribution of AI technology impact five wide-reaching areas: appropriate transparency and explainability; accountability and governance; safety, security and robustness; fairness; and contestability and redress. Overall, the review aims to learn how AI foundation models can, and likely will, impact both competition and consumer protections. Regulating bodies tasked with finding the answers include the Competition and Markets Authority (CMA), which helps people and businesses in competitive markets while working against unethical practices. "It's crucial that the potential benefits of this transformative technology are readily accessible to UK businesses and consumers while people remain protected from issues like false or misleading information," Sarah Cardell, CMA's chief executive, said in a statement.

Slack has become the latest developer to commit to integrating AI inside of its platform, with a new Slack GPT integration that will use chatbots as your own personal business assistants. But Slack GPT also highlights another AI quirk that may fly under the radar: its short-term attention span. In March, Salesforce's Slack division announced a ChatGPT app, allowing the OpenAI chatbot to be pulled into the Slack conversational interface. Now, Slack is expanding that: Slack GPT will support ChatGPT, the Claude chatbot, and Slack's own eventual Einstein GPT chatbot as well. To avoid forcing users to scroll back through reams of chat text (say, after returning from vacation) Slack GPT will simply summarize it all for you.

Google's acceleration comes as a cacophony of voices -- including notable company alumnae and industry veterans -- are calling for the AI developers to slow down, warning that the tech is developing faster than even its inventors anticipated. Geoffrey Hinton, one of the pioneers of AI tech who joined Google in 2013 and recently left the company, has since gone on a media blitz warning about the dangers of supersmart AI escaping human control. Pichai, along with the CEOs of OpenAI and Microsoft, will meet with White House officials on Thursday, part of the administration's ongoing effort to signal progress amid public concern, as regulators around the world discuss new rules around the technology.

The Biden administration's investment in responsible AI research and development is a $140 million grant, which will increase the number of national AI research institutes. These institutes are focused on advancing artificial intelligence research in areas ranging from public health to cybersecurity. The investment is just a fraction of the billions that private sector companies are pouring into advancing the technology. Microsoft previously invested $10 billion in OpenAI.

At its World Tour NYC event, Salesforce has introduced Slack GPT, which it describes as a three-pronged vision that integrates AI features into the business messaging app. Slack GPT is comprised of AI-powered features built natively into the app, a new AI-ready platform that was recently made available to developers, and the availability of Einstein GPT in the app that will power its ability to instantly generate insights and summaries. Einstein GPT was developed by Salesforce as a generative AI for customer relationship management (CRM) and could assist businesses with tasks related to sales. The integrated AI features will give users access to a workflow builder that doesn't require them to know how to code. In it, they can automatically create or update a canvas, which is Slack's tool designed for collaboration.

Ahead of a meeting between Vice President Kamala Harris and the heads of America's four leading AI tech companies -- Alphabet, OpenAI, Anthropic and Microsoft -- the Biden Administration announced Thursday a sweeping series of planned actions to help mitigate some of the risks that these emerging technologies pose to the American public. That includes $140 million to launch seven new AI R&D centers as part of the National Science Foundation, extracting commitments from leading AI companies to participate in a "public evaluation" of their AI systems at DEFCON 31, and ordering the Office of Management and Budget (OMB) to draft policy guidance for federal employees. "The Biden Harris administration has been leading on these issues since long before these newest generative AI products debuted last fall," a senior administration official said during a reporters call Wednesday. The Administration unveiled its AI Bill of Rights "blueprint" last October, which sought to "help guide the design, development, and deployment of artificial intelligence (AI) and other automated systems so that they protect the rights of the American public," per a White House press release. "At a time of rapid innovation, it is essential that we make clear the values we must advance, and the common sense we must protect," the administration official continued.

ChatGPT has stoked new hopes about the potential of artificial intelligence--but also new fears. Today the White House joined the chorus of concern, announcing it will support a mass hacking exercise at the Defcon security conference this summer to probe generative AI systems from companies including Google. The White House Office of Science and Technology Policy also said that $140 million will be diverted towards launching seven new National AI Research Institutes focused on developing ethical, transformative AI for the public good, bringing the total number to 25 nationwide. The announcement came hours before a meeting on the opportunities and risks presented by AI between vice president Kamala Harris and executives from Google and Microsoft as well as the startups Anthropic and OpenAI, which created ChatGPT. The White House AI intervention comes as appetite for regulating the technology is growing around the world, fueled by the hype and investment sparked by ChatGPT.

Bing AI is now open to all--sort of. Three months after debuting its revamped search engine, Microsoft has announced that it's now moving into open preview. You'll still need to sign into Bing on the Edge browser (or the Bing mobile apps) to use the chatbot, but at least you no longer have to deal with a waitlist. As if to celebrate this new phase of Bing (powered by OpenAI's GPT-4), Microsoft is also rolling out several new features. For one, it can go beyond mere text responses to deliver charts, graphs and rich formatting.

Deep generative models such as GANs, normalizing flows, and diffusion models are powerful regularizers for inverse problems. They exhibit great potential for helping reduce ill-posedness and attain high-quality results. However, the latent tensors of such deep generative models can fall out of the desired high-dimensional standard Gaussian distribution during inversion, particularly in the presence of data noise and inaccurate forward models, leading to low-fidelity solutions. To address this issue, we propose to reparameterize and Gaussianize the latent tensors using novel differentiable data-dependent layers wherein custom operators are defined by solving optimization problems. These proposed layers constrain inverse problems to obtain high-fidelity in-distribution solutions. We validate our technique on three inversion tasks: compressive-sensing MRI, image deblurring, and eikonal tomography (a nonlinear PDE-constrained inverse problem) using two representative deep generative models: StyleGAN2 and Glow. Our approach achieves state-of-the-art performance in terms of accuracy and consistency.