Rocky Start for Anthropic’s Claude Code

Launch Issues with Auto-Update Function

The launch of Anthropic’s coding tool, Claude Code, has gotten off to a rocky start. According to reports on GitHub, the auto-update function contained buggy commands that rendered some workstations unstable and broken.

Buggy Commands Cause Problems

When installed at the "root" or "superuser" levels – permissions that give programs the ability to make operating system-level changes – the buggy commands would let applications modify typically restricted file directories and, in the worst-case scenario, "brick" systems.

File Permissions Issues

The problematic Claude Code auto-update commands changed the access permissions of certain critical system files. Permissions define which programs and users can read or modify files, or run certain apps. One GitHub user reported that they were forced to employ a "rescue instance" to fix the permissions of files Claude Code’s commands inadvertently broke.

Response from Anthropic

Anthropic has since removed the problematic commands from Claude Code and added a link in the program directing users to a troubleshooting guide. However, the link initially had a typo – but Anthropic has since fixed this issue.

Conclusion

The launch of Claude Code has been marred by technical issues, causing problems for some users. While Anthropic has taken steps to address the problem, the rocky start may raise concerns about the tool’s reliability and stability.

Frequently Asked Questions

Q: What were the problems with Claude Code’s auto-update function?
A: The auto-update function contained buggy commands that rendered some workstations unstable and broken.

Q: What were the consequences of installing Claude Code at the "root" or "superuser" levels?
A: Installing Claude Code at these levels allowed applications to modify typically restricted file directories and, in the worst-case scenario, "brick" systems.

Q: What did Anthropic do to address the issue?
A: Anthropic removed the problematic commands from Claude Code and added a link in the program directing users to a troubleshooting guide.

Q: Did the link to the troubleshooting guide initially have a typo?
A: Yes, the link initially had a typo, but Anthropic has since fixed this issue.

Brazil’s Antitrust Regulator Orders Apple to Allow Alternative In-App Payment Options and Sideloading

The judge who issued today’s ruling said that “Apple has already complied with similar obligations in other countries without demonstrating significant impact or irreparable harm to its economic model,” Valor reports.

Background on the Investigation

Brazil’s investigation into Apple began with a complaint filed by Mercado Livre in 2022. In November, Brazil gave Apple 20 days to let developers offer alternative in-app payment options and allow sideloading, but in early December, the injunction was overturned.

Reaction from Apple

The company plans to appeal the ruling, according to a statement shared with The Verge.

“Apple believes in vibrant and competitive markets where innovation can flourish,” spokesperson Fred Sainz says. “We face competition in every segment and jurisdiction where we operate, and our focus is always the trust of our users. We are concerned that the interim measures proposed by CADE would undermine the privacy and security of our users and intend to appeal today’s decision.”

Conclusion

In conclusion, the Brazilian antitrust regulator has ordered Apple to allow alternative in-app payment options and sideloading, but the company plans to appeal the decision. This move is seen as a significant development in the ongoing debate over the dominance of Apple in the digital payments market.

FAQs

Q: What is the background on the investigation?

A: The investigation into Apple began with a complaint filed by Mercado Livre in 2022.

Q: What is the ruling and what does it require of Apple?

A: The ruling requires Apple to allow alternative in-app payment options and sideloading.

Q: How has Apple responded to the ruling?

A: Apple plans to appeal the ruling, citing concerns over the impact on user privacy and security.

Q: What is the next step for this case?

A: The case is likely to proceed to an appeal, with the final decision yet to be made.

Imagine that a robot is helping you clean the dishes. You ask it to grab a soapy bowl out of the sink, but its gripper slightly misses the mark.

Using a new framework developed by MIT and NVIDIA researchers, you could correct that robot’s behavior with simple interactions. The method would allow you to point to the bowl or trace a trajectory to it on a screen, or simply give the robot’s arm a nudge in the right direction.

Unlike other methods for correcting robot behavior, this technique does not require users to collect new data and retrain the machine-learning model that powers the robot’s brain. It enables a robot to use intuitive, real-time human feedback to choose a feasible action sequence that gets as close as possible to satisfying the user’s intent.

When the researchers tested their framework, its success rate was 21 percent higher than an alternative method that did not leverage human interventions.

In the long run, this framework could enable a user to more easily guide a factory-trained robot to perform a wide variety of household tasks even though the robot has never seen their home or the objects in it.

“We can’t expect laypeople to perform data collection and fine-tune a neural network model. The consumer will expect the robot to work right out of the box, and if it doesn’t, they would want an intuitive mechanism to customize it. That is the challenge we tackled in this work,” says Felix Yanwei Wang, an electrical engineering and computer science (EECS) graduate student and lead author of a paper on this method.

His co-authors include Lirui Wang PhD ’24 and Yilun Du PhD ’24; senior author Julie Shah, an MIT professor of aeronautics and astronautics and the director of the Interactive Robotics Group in the Computer Science and Artificial Intelligence Laboratory (CSAIL); as well as Balakumar Sundaralingam, Xuning Yang, Yu-Wei Chao, Claudia Perez-D’Arpino PhD ’19, and Dieter Fox of NVIDIA. The research will be presented at the International Conference on Robots and Automation.

Mitigating misalignment

Recently, researchers have begun using pre-trained generative AI models to learn a “policy,” or a set of rules, that a robot follows to complete an action. Generative models can solve multiple complex tasks.

During training, the model only sees feasible robot motions, so it learns to generate valid trajectories for the robot to follow.

While these trajectories are valid, that doesn’t mean they always align with a user’s intent in the real world. The robot might have been trained to grab boxes off a shelf without knocking them over, but it could fail to reach the box on top of someone’s bookshelf if the shelf is oriented differently than those it saw in training.

To overcome these failures, engineers typically collect data demonstrating the new task and re-train the generative model, a costly and time-consuming process that requires machine-learning expertise.

Instead, the MIT researchers wanted to allow users to steer the robot’s behavior during deployment when it makes a mistake.

But if a human interacts with the robot to correct its behavior, that could inadvertently cause the generative model to choose an invalid action. It might reach the box the user wants, but knock books off the shelf in the process.

“We want to allow the user to interact with the robot without introducing those kinds of mistakes, so we get a behavior that is much more aligned with user intent during deployment, but that is also valid and feasible,” Wang says.

Their framework accomplishes this by providing the user with three intuitive ways to correct the robot’s behavior, each of which offers certain advantages.

First, the user can point to the object they want the robot to manipulate in an interface that shows its camera view. Second, they can trace a trajectory in that interface, allowing them to specify how they want the robot to reach the object. Third, they can physically move the robot’s arm in the direction they want it to follow.

“When you are mapping a 2D image of the environment to actions in a 3D space, some information is lost. Physically nudging the robot is the most direct way to specifying user intent without losing any of the information,” says Wang.

Sampling for success

To ensure these interactions don’t cause the robot to choose an invalid action, such as colliding with other objects, the researchers use a specific sampling procedure. This technique lets the model choose an action from the set of valid actions that most closely aligns with the user’s goal.

“Rather than just imposing the user’s will, we give the robot an idea of what the user intends but let the sampling procedure oscillate around its own set of learned behaviors,” Wang explains.

This sampling method enabled the researchers’ framework to outperform the other methods they compared it to during simulations and experiments with a real robot arm in a toy kitchen.

While their method might not always complete the task right away, it offers users the advantage of being able to immediately correct the robot if they see it doing something wrong, rather than waiting for it to finish and then giving it new instructions.

Moreover, after a user nudges the robot a few times until it picks up the correct bowl, it could log that corrective action and incorporate it into its behavior through future training. Then, the next day, the robot could pick up the correct bowl without needing a nudge.

“But the key to that continuous improvement is having a way for the user to interact with the robot, which is what we have shown here,” Wang says.

In the future, the researchers want to boost the speed of the sampling procedure while maintaining or improving its performance. They also want to experiment with robot policy generation in novel environments.

StubHub Considers Going Public with $1 Billion IPO

Talks with Bankers Intensify Ahead of Potential Listing

StubHub, a ticketing software company, has held discussions with bankers in recent weeks about holding an initial public offering (IPO) before the end of the year, according to two sources with knowledge of the matter. The talks are preliminary, and the company has not yet made a formal decision to go public.

Growing Appetite for Tech IPOs

The tech public offering market has gotten off to a slow start this year, but a growing number of companies appear ready to contemplate public listings, despite market volatility stemming from President Trump’s constant changes to economic policies. CoreWeave, a provider of cloud computing services for AI companies, has unveiled its investor prospectus, while Discord, the social chat app popular with the video game community, has met with investment bankers to discuss preparations for an IPO.

StubHub’s History and Growth

StubHub was founded in 2000 by Eric Baker and Jeff Fluhr and quickly grew in its early days. The company allowed users to buy secondary-market tickets for various events, including concerts and sports. In 2006, Baker founded ViaGogo, a similar ticketing company focused on European ticket reselling and events. In 2019, ViaGogo announced it was buying StubHub for $4 billion. The company faced significant challenges during the COVID-19 pandemic, with lockdowns causing a 95% decline in its business. However, the company saw a surge in business as lockdowns lifted and fans returned to venues.

IPO Plans Shelved Last Year

The company eyed an IPO last year at a $16.5 billion valuation, but shelved its plans amid a dearth of public listings of similar companies. However, executives at StubHub have said that the company’s business is still growing, with gross ticket sales increasing by nearly 50% in the last three months of 2024.

Recent Developments and Concerns

Recently, there have been concerns on Wall Street that ticketing companies are seeing slower sales growth as fans cool on concerts after an initial surge of post-pandemic interest. However, StubHub’s executives have attributed this decline to the company’s ability to adapt to changing market conditions and focus on new revenue streams.

Conclusion

StubHub’s potential IPO is a significant development in the tech industry, which has seen a slow start to the year. The company’s growth and adaptability have allowed it to weather the challenges posed by the pandemic, and its plans for an IPO could be a major boost for the ticketing industry.

Frequently Asked Questions

Q: What is StubHub’s current valuation?
A: StubHub’s current valuation is not publicly disclosed, but the company is aiming to raise over $1 billion in its potential IPO.

Q: Who are the founders of StubHub?
A: StubHub was founded by Eric Baker and Jeff Fluhr in 2000.

Q: What are the current challenges facing StubHub?
A: The company is facing slower sales growth in the ticketing industry, but its executives attribute this to their ability to adapt to changing market conditions and focus on new revenue streams.

Q: What is the current state of the tech IPO market?
A: The tech public offering market has gotten off to a slow start this year, but a growing number of companies appear ready to contemplate public listings, despite market volatility.