Online Image Abuse: A Persistent Problem

Challenges in Reporting and Takedown Processes

At the start of last August, Point de Contact told WIRED that only two images on four different Microsoft servers remained. "We deeply regret that this issue took almost 10 months of communication between the victim, Microsoft, and us as an NGO to be resolved," the NGO said in an email at the time.

Microsoft’s Response and Efforts to Improve

Microsoft digital safety chief Gregoire says Liu’s situation has spurred her team to try to improve reporting processes and relationships with victim aid groups. Point de Contact initially flagged links over which the company didn’t have control, according to Gregoire. She declined to elaborate on the circumstances. Dirani says this explanation was never communicated to him, and it remains unclear why the links were not "actionable."

Strained Relationships with Tech Giants

Only after Powell cornered Thomas over Liu’s case did Microsoft obtain the URLs upon which it could act. "We’re thankful, to be perfectly honest, to the spontaneous connection at TrustCon," Gregoire says. But it shouldn’t be needed again: Point de Contact now has a more direct way to stay in touch, she says.

Other victim aid groups say their relationships with tech giants remain challenging. Last year, a WIRED investigation revealed that executives at Google rejected numerous ideas raised by staff and outside advocates that aimed to proactively counter access to problematic imagery in search results. Some survivors have found that the fastest way to get content removed is by filing copyright claims, a tactic those working in the online safety industry say is inadequate.

Consistency in Policies and Processes

The lack of consistency in policies and processes among tech companies contributes to delays in securing takedowns, according to Emma Pickering, the head of technology facilitated abuse at Refuge, the UK’s largest domestic abuse organization. "They all just respond however they choose to—and the response usually is incredibly poor," she says. (Google introduced new policies in July 2024 to accelerate removals.)

Microsoft’s Particular Challenges

Pickering claims Microsoft, in particular, has been difficult. "I’ve recently been told if I want to engage with them, we need to provide evidence that we use their platform and we promote them," she says, adding Refuge is trying to engage with as many tech platforms as possible.

Microsoft’s Efforts to Stem the Need for Takedowns

Microsoft’s Gregoire says she will look into these concerns and is open to dialogue. The company hopes to stem the need for takedowns, in part, by scaring off perpetrators. This past December, Microsoft sued a group of 10 unknown individuals who allegedly circumvented safeguards on Azure and used an AI tool to generate offensive images, including some Gregoire described as sexually harmful. "We don’t want our services to be abused to cause harm," she says.

Conclusion

The challenges faced by Liu and other survivors of online image abuse are a stark reminder of the need for improved reporting processes and relationships between tech giants and victim aid groups. While Microsoft’s efforts to improve are a step in the right direction, more needs to be done to address the lack of consistency in policies and processes.

FAQs

Q: What is the current state of online image abuse?
A: Online image abuse remains a persistent problem, with many survivors struggling to have explicit images removed from the internet.

Q: How do tech companies handle reports of online image abuse?
A: The process is often inconsistent and plagued by delays, with some companies rejecting proactive solutions proposed by staff and outside advocates.

Q: What is being done to address the issue?
A: Microsoft is working to improve reporting processes and relationships with victim aid groups, while a proposed US bill aims to require websites to remove unwanted explicit images within 48 hours.

Q: What resources are available for those affected by online image abuse?
A: The National Suicide Prevention Lifeline (1-800-273-8255) and the Crisis Text Line (text HOME to 741741) are available for support. International associations can be found at the International Association for Suicide Prevention.

Figure Unveils New Machine Learning Model for Humanoid Robots

Introducing Helix: A Generalist Vision-Language-Action (VLA) Model

Figure, a Bay Area robotics firm, has announced a new machine learning model for humanoid robots, called Helix. The model is a generalist Vision-Language-Action (VLA) model, which is a new phenomenon in robotics. VLA models leverage vision and language commands to process information, and Helix is designed to work in real-time, combining visual data and language prompts to control a robot.

How Helix Works

Helix is capable of picking up thousands of novel household items with varying shapes, sizes, colors, and material properties, simply by asking in natural language. The platform is designed to bridge the gap between vision and language processing, allowing a robot to visually assess its environment and then perform a task.

Real-World Applications

Figure is showcasing the VLM by highlighting its work with its 02 humanoid robot in the home environment. The company is demonstrating the potential of Helix by highlighting its ability to control two robots working together to perform various household tasks.

Challenges and Opportunities

The home is a challenging and complex setting for testing these sorts of training models. Teaching robots to do complex tasks in the kitchen, for example, opens them up to a broad range of actions in different settings. However, manual programming won’t scale for the home, and training requires hundreds of hours of repetition.

Conclusion

Helix is a significant step forward in the development of humanoid robots, enabling them to perform complex tasks in a variety of environments. With its ability to work in real-time and control multiple robots, Helix has the potential to revolutionize the way we live and work.

Frequently Asked Questions

Q: What is Helix?
A: Helix is a generalist Vision-Language-Action (VLA) model that combines visual data and language prompts to control a robot in real-time.

Q: How does Helix work?
A: Helix works by visually assessing its environment and then performing a task based on language prompts.

Q: What are the real-world applications of Helix?
A: Helix has the potential to revolutionize the way we live and work, enabling robots to perform complex tasks in a variety of environments.

Q: What are the challenges associated with developing Helix?
A: The home is a challenging and complex setting for testing these sorts of training models, and manual programming won’t scale for the home.

Smart Home Data Privacy: Skylight Calendar

Spring is my busy season. Since my kids were old enough to play team sports, the impending arrival of March brings an avalanche of organizational tasks. Practice schedules, game schedules, and snack signups equal so many dates to remember. They arrive in different ways — paper flyers, Excel sheets, text chains, or through yet another team sports app — but somehow never in a Google Calendar link. It’s a trying time.

But this year, I have a new assistant: Sidekick, an AI-powered planning feature for the Skylight Calendar. I just forward every lengthy email to my Sidekick and upload pictures of schedules and spreadsheets to its app. From there, the AI parses the data, packages it into events, and pushes those to my Google Calendar and to my family’s calendars on the big, bright, 27-inch Skylight Calendar Max smart display that’s mounted by our breakfast counter for all to see.

I’m now prepared to tackle two high school sports seasons, along with all the other scheduling chaos that comes with being a busy working parent to two busy teens. It has taken a decade and significant advances in AI to get here, but parents, this is a game-changer.

The Skylight Calendar Max

The Skylight Calendar Max next to the Echo Show 21. The Show can be wall-mounted or tabletop-mounted, the Cal Max can only go on the wall. The Cal Max has a better calendar interface.

Smart Home Data Privacy

It’s important to note that when you use Skylight, the company has access to any data you store on the device. Skylight says it follows all industry best practices to protect that data.

“[O]ur systems do need to ‘see’ that data, but we follow industry best practices in terms of customer privacy, including but not limited to:

• Encrypting all data in transit and at rest
• Explicitly requesting permission from customers to access their data if their support request requires it
• Using best-in-class systems for storing and cycling those encryption keys
• Never storing data that we don’t need
• Never using that data for any purpose other than serving the customer
• Allowing customers to delete all of their data from our database with a few taps
• Regularly hiring third-party auditors to review our code and infrastructure for vulnerabilities”

Competitors

Another more direct competitor to the Skylight Calendar is the Hearth display, which I haven’t tested. It’s the same size as the Skylight, but at $699, it’s more expensive. It does have Hearth Helper, an AI assistant that can import events from flyers and such if you send a photo of it via text. This costs $9 a month ($86.40 a year), almost double Skylight’s fee.

Conclusion

Sidekick brings another level of utility to the Skylight Calendar’s already good hardware. Yes, it’s a big screen running Android, and yes, Gemini and ChatGPT et al. could surpass Sidekick’s schedule parsing abilities one day soon. Yes, it is dependent on Wi-Fi, and if the company goes out of business, you may be left with a big, defunct screen. Skylight tells me if this happened, it “would find a creative way to support all existing products out in the wild.”

But the Skylight team has created a compelling product for corralling the chaos of family schedules. If that’s something you need, and you don’t want to spend time creating a DIY version with a monitor powered by a Raspberry Pi and running DAKboard (as one Verge commenter suggested), this is an out-of-the-box solution that works very well.

FAQs

Q: What is the cost of the Skylight Calendar?
A: The cost of the Skylight Calendar Max is $399.

Q: How does the Skylight Calendar work?
A: The Skylight Calendar works by importing data from various sources, such as email, spreadsheets, and text messages, and then displaying that data on the calendar.

Q: What is Sidekick?
A: Sidekick is an AI-powered planning feature for the Skylight Calendar that helps to parse data and package it into events.

Q: Is the Skylight Calendar secure?
A: Yes, the Skylight Calendar is secure, as the company follows all industry best practices to protect customer data.

Q: Are there any competitors to the Skylight Calendar?
A: Yes, there are competitors to the Skylight Calendar, such as the Hearth display, which has a similar feature set but is more expensive.