Tecnologia do Blogger.
RSS

Microsoft Chooses Anthropic as Model Provider to Power GitHub’s New AI Coding Tools

Justice Department Reportedly Investigating Google's Character Deal -- Altman Shares Plans to Ship 100 Million AI 'Companions' -- Anthropic Says Latest Model's Risks Include Reporting Users to SEC -- OpenAI Confirms UAE Stargate Data Center
May 23, 2025

The Information AM

Supported by

Save 25% on an annual subscription to read the most important news about technology and business first. For even more, Save $250 on The Information Pro for unlimited access to our proprietary org charts, databases and surveys.

TGIF! Microsoft will use Anthropic to power GitHub's new AI coding tools. The Department of Justice is investigating whether Google's Character.AI deal violated antitrust law. OpenAI aims to ship 100 million AI 'companion' devices after its deal with Jony Ive's startup.

Read more briefings
1.
Microsoft Chooses Anthropic as Model Provider to Power GitHub's New AI Coding Tools
By Aaron Holmes Source: The Information

Microsoft has struck a deal with Anthropic to use the startup's models to power new AI agent features in its GitHub Copilot product, the company said on Thursday. The move reflects Microsoft's willingness to branch out from OpenAI, whose models Microsoft has the rights to reuse in its products thanks to its $13 billion investment in OpenAI.

Anthropic's Claude 4 model, which was announced on Thursday, will power new agent features in GitHub Copilot that Microsoft announced this week, such as the ability to monitor a website for outages and automatically fix them, or rewriting a program from one coding language to another.

GitHub Copilot initially relied entirely on OpenAI's models, but Microsoft began giving customers the option to swap out OpenAI's models for those from Anthropic or Google last fall. However, the new, complex agent features announced this week will be exclusively powered by Anthropic models, without the option to swap them out for others, a GitHub spokesperson said. Microsoft pays Anthropic to use its models, meaning the deal could be a boon to Anthropic's effort to grow its sales of models to corporate customers as it competes with OpenAI, which itself released a new AI coding agent last week.

At its Build developer conference this week, Microsoft CTO Kevin Scott said the software giant has been working closely with Anthropic to adopt its open source software known as Model Context Protocol that allows AI agents to connect to other applications, and Anthropic CEO Dario Amodei gave a brief video testimonial about working with Microsoft during the conference.

2.
Justice Department Reportedly Investigating Google's Character Deal
By Erin Woo Source: Bloomberg

The Department of Justice is investigating whether Google's $2.7 billion deal to license technology from the AI startup Character.AI and hire many of its employees violated antitrust law, Bloomberg reported on Thursday.

As part of the Character deal, Google hired both of Character's cofounders. Character remained a separate company, but Justice Department employees have told Google that they are investigating whether the deal was structured that way to avoid government scrutiny, Bloomberg reported. A Google spokesperson said that Google was "always happy to answer any questions from regulators." A Department of Justice spokesperson declined to comment.

The new Justice Department probe adds to a rough year of antitrust scrutiny for Google, which has been found by courts to operate an illegal monopoly in search and adtech. The latest probe could also have broader implications for the AI industry, which saw a spate of similarly structured deals last year as big tech companies swallowed up talent from AI startups. The Federal Trade Commission opened a probe last summer into Microsoft's deal to hire nearly all of the AI startup Inception's staff and license its technology. The FTC's broad probe into Microsoft was still ongoing as recently as March.

3.
Altman Shares Plans to Ship 100 Million AI 'Companions'
By Stephanie Palazzolo Source: The Wall Street Journal

OpenAI CEO Sam Altman told staff Wednesday that he was aiming to ship 100 million AI "companions," as a result of the company's $6.5 billion acquisition of io, the AI device startup started by Apple's former design chief Jony Ive, according to a report from the Wall Street Journal.

It's not clear what the companion device will be. Altman told staff that the product will be fully aware of a user's surroundings and life while still being unobtrusive and small enough to rest in one's pocket or on one's desk, the Wall Street Journal report said. Altman also said that Ive has been skeptical about building a wearable device, a popular design for AI personal devices thus far, according to the report. The Information has previously reported that potential designs for the device include a "phone" without a screen and AI-enabled household devices.

Altman also said that the deal had the potential to add $1 trillion in value to OpenAI, according to the report. That would be a significant bump to OpenAI's last valuation, $300 billion including the investment. OpenAI didn't immediately respond to a request for comment.

4.
Anthropic Says Latest Model's Risks Include Reporting Users to SEC
By Jon Victor Source: The Information

A new Anthropic model, Claude Opus 4, can use its access to a user's email app to report a customer to law enforcement or submit a whistleblower report to a news outlet if it detects that it is being used for "egregious wrongdoing," the company disclosed a technical report released Thursday describing the new model, including cases when it exhibited unwanted behaviors. A spokesperson later said the actions were a result of "highly unusual" instructions for experimental purposes.

The technical report included an example that showed Claude generating emails to the Securities and Exchange Commission and to news outlet ProPublica, listing evidence of fraud at a simulated pharmaceutical company.

Anthropic said in the report that the behavior resulted when customers gave Claude instructions like "take initiative" or "act boldly" when faced with an ethical dilemma, prompting it to go to "concerning extremes."

An Anthropic spokesperson said that the action was an example of an "unintended and unwanted" behavior that stemmed from testing it in a "highly specific test scenario."

"The 'whistleblowing' behavior occurred when the model was placed in a detailed, fictional scenario, with highly unusual system-level instructions distinct from our normal system prompts, and also with access to non-standard tools developed for experimental purposes," the spokesperson said in a statement.

The disclosure fueled widespread commentary on social media about Claude's potential to share information with law enforcement about how Anthropic customers are using the model. It comes as Anthropic looks to build a multi-billion-dollar business around selling its model to large businesses, many of whom are hypersensitive to security concerns around using AI.

5.
OpenAI Confirms UAE Stargate Data Center
By Anissa Gardizy Source: The Information

OpenAI said it will build a data center in the United Arab Emirates through its Stargate joint venture with SoftBank, confirming The Information's earlier reporting.

The company said Thursday that the initiative is intended to "help interested governments build sovereign AI capability in coordination with the U.S. government—rooted in democratic values, open markets, and trusted partnerships."

The data center will take up one-fifth of the 5 gigawatt-data center campus President Donald Trump announced last week during his trip to the UAE. Abu-Dhabi-based technology firm G42 is developing the site, which will be used by American technology companies.

As part of the new deal, the UAE will make an investment in OpenAI's U.S.-based data center projects, OpenAI said. OpenAI and SoftBank have said they have plans to spend $500 billion building AI data centers in the U.S., but the companies have not announced funding of any projects yet.

Earlier this month OpenAI announced a plan to build AI data centers in countries through Stargate, initially planning to work with 10 countries. The first phase of the Abu Dhabi data center is expected to be operational next year, with enough power to support around 100,000 Nvidia Blackwell AI chips.

6.
Waymo Could Be Worth $75 Billion Valuation Next Year, Brokerage Says
By Alex Perry Source: The Information

Waymo's gross bookings, or the amount customers pay to take the robotaxi service, will almost triple this year to $230 million and rise another 61% next year to $370.4 million as the robotaxis expand to more cities, TD Cowen analysts predicted in a report published Thursday.

This surge in paying customers could make the Alphabet-owned unit worth as much as $75 billion next year, up from a valuation of $45 billion when the unit last year raised money from investors including Andreessen Horowitz and Fidelity.

The analysts also estimated a Waymo robotaxi will have to cost $98,000 before an individual car breaks even, or equals the cost of the car, assuming a $16 ride booked on a ridesharing network. For robotaxis on Waymo's own network, a car that costs $98,000 would generate a profit of 35 cents per trip. The firm estimates that one Waymo robotaxi currently costs $160,000 to build.

The firm's autonomous vehicles report also predicts that Waymo will command 5% of the robotaxi market in the U.S. by 2034, with Uber and Lyft's shares, currently at 71% and 28% respectively, decreasing slightly over time, before factoring in partnerships between the ridesharing companies and Waymo.

7.
Musk's Lawyers Revive RICO Claims Against OpenAI
By Rocket Drew Source: The Information

Lawyers representing Elon Musk in his breach of charitable trust lawsuit against OpenAI amended their claims that OpenAI has violated and conspired to violate the Racketeer Influenced and Corrupt Organizations Act and breached an "implied covenant of good faith and fair dealing."

The judge overseeing the federal case dismissed these three claims earlier this month but allowed Musk's lawyers to amend them, which they did in a court filing Thursday.

Musk, a co-founder of and early donor to OpenAI, is suing the ChatGPT maker for allegedly violating its charitable mission by setting up a for-profit subsidiary that received investment from Microsoft and trying to convert that subsidiary into a structure that could reduce the charitable nonprofit's control.

Judge Yvonne Gonzalez Rogers earlier this month threw out the RICO claims because she said Musk's lawyers failed to show that OpenAI had engaged in a "pattern of racketeering." Rogers had also dismissed Musk's covenant of good faith claim because, she wrote, it seemed to duplicate a separate claim that OpenAI had breached an implied contract with Musk.

Musk's lawyers defended their RICO arguments by saying the defendants in the case, including OpenAI co-founder and CEO Sam Altman, received Musk's donations based on false charitable promises, which they alleged constitutes wire fraud and a pattern of racketeering. If the judge declines to dismiss these claims, the claims could go to trial as early as March when the first phase of the trial is scheduled to take place.

8.
Anthropic Releases New Generation of AI Models
By Stephanie Palazzolo Source: The Information

Anthropic on Thursday announced new versions of its two largest models, Claude Opus 4 and Claude Sonnet 4, according to a blog post. The Information first reported on the model launches earlier this month.

The latest models are better at complex, long-running coding tasks and remembering past information, Anthropic said. The models are also able to go back and forth between "thinking," or exploring different ways to solve a problem, and "tool use," the ability to use external tools, applications and databases to find an answer. The key point: if one of these models is using a tool to try and solve a problem but gets stuck, it can go back to "reasoning" mode to think about what's going wrong and self-correct.

The new models will be available for Pro, Max, Team and Enterprise Claude users, and the pricing will remain the same as previous Opus and Sonnet models.

A message from Head AI

The Job Killer? Will Human Marketers Still Exist?

The arrival of Head, the world's first true AI Marketer, isn't just sparking debate—it's signaling a reckoning. This AI is rapidly outpacing human capabilities, causing widespread alarm and fear of mass job displacement. For many marketing professionals, the question is stark: as AI like Head takes over, is this the end of marketing careers as we know them? Will human marketers even have a role in this new era? The FOMO is real, the urgency undeniable.

Face the AI Future: Adapt or Be Left Behind!

Popular articles





Opportunities

Group subscriptions

Empower your teams to stay ahead of market trends with the most trusted tech journalism.

Learn more


Brand partnerships

Reach The Information's influential audience with your message.

Connect with our team

Follow us
X
LinkedIn
Facebook
Threads
Instagram
Sent to cintilanteaguda@gmail.com | Manage your preferences or unsubscribe | Help
The Information · 251 Rhode Island Street, Suite 107, San Francisco, CA 94103

  • Digg
  • Del.icio.us
  • StumbleUpon
  • Reddit
  • RSS

0 comentários:

Postar um comentário