Facebook is requesting access to users' camera rolls to suggest AI-edited versions of photos, including those not uploaded to Facebook.
Users are prompted to opt into 'cloud processing' when creating a Story, allowing Facebook to upload media to its servers for AI-generated suggestions like collages and recaps.
Meta's AI Terms state that by allowing access, users agree to let Meta analyze their media and facial features for AI processing, though the media isn't used for ad targeting.
The feature raises privacy concerns as Meta's AI can retain and use personal information from photos to personalize AI outputs, with unclear definitions of what constitutes personal data.
Users have found the feature in Facebook's Settings under 'Camera roll sharing suggestions,' where they can toggle cloud processing on or off.
Meta's AI terms, enforceable since June 23, 2024, allow the company to review interactions with its AIs, including conversations, potentially involving human reviewers.
The feature has sparked minimal backlash, with some users questioning its implications, but Meta has not responded to requests for comment.
The US Copyright Office is facing unprecedented turbulence amid the AI boom and leadership disputes.
Shira Perlmutter, the former Copyright Register, was abruptly fired by the White House and is now suing the Trump administration, claiming her dismissal was invalid.
The Copyright Office currently lacks a clear leader, raising concerns about the validity of its operations and copyright certificates.
The firing of Perlmutter followed the dismissal of Librarian of Congress Carla Hayden, leading to confusion over who holds authority in the Library of Congress.
Paul Perkins and Brian Nieves were appointed to roles in the Copyright Office but have not assumed their duties, leaving the office in limbo.
Congress is negotiating with the White House to resolve the leadership dispute, with some lawmakers supporting Perlmutter's claim to her position.
The Copyright Office paused issuing registration certificates after Perlmutter's dismissal, resuming later without her signature, sparking legal concerns.
Experts debate whether unsigned copyright registrations are legally vulnerable, as the law does not explicitly require the Register's signature.
The absence of a Register impacts other functions, such as recertifying organizations like the Mechanical Licensing Collective (MLC) and appointing members to the Copyright Claims Board.
The Copyright Office's instability comes at a critical time, as its recent report on AI and fair use is being cited in lawsuits despite not being finalized.
Meta invests $14 billion in Scale AI, acquiring ~49% of the company, effectively making it a quasi-acquisition.
The deal is similar to other tech giant investments aimed at poaching top talent, leaving the original company as a shell.
Scale AI's core business of data labeling doesn't align with Meta's consumer social media focus, suggesting the investment is primarily to acquire Alexandr Wang.
Meta is struggling in the AI race, lagging behind competitors like OpenAI, Google, and Anthropic in both consumer and developer markets.
Meta's internal culture, marked by political infighting and poor leadership under Yann LeCun, has hindered its AI progress.
Zuckerberg is aggressively recruiting top AI talent, offering massive salaries to build a 'super intelligence' team, but this strategy may be misguided.
AI success often stems from collaborative environments rather than individual superstars, making Meta's focus on poaching top talent questionable.
Despite challenges, Zuckerberg's leadership and vision give hope that Meta can turn its AI efforts around.
Douglas Hofstadter discusses his identity as an artist, emphasizing his pursuit of beauty through various forms of art, including ambigrams, piano compositions, and poetry translations.
Hofstadter views beauty as a supreme goal in life, separate from truth, which can sometimes be ugly, as exemplified by current events like the devastation in Gaza.
The concept of 'strange loops' is central to Hofstadter's work, particularly in explaining consciousness and the self, though he declines to summarize it briefly here.
Hofstadter expresses skepticism about free will, describing human actions as the result of internal battles between conflicting desires rather than free choice.
He considers consciousness an 'illusion,' a concept he elaborates on in his book 'I Am a Strange Loop.'
Hofstadter is deeply concerned about the rapid advancement of AI, likening humanity's current trajectory to driving recklessly into a fog bank.
He entertains the possibility of a technological Singularity but finds the idea frightening, questioning whether future AI could become humanity's evolutionary successors.
Hofstadter dismisses the simulation hypothesis and the concept of God as unserious, comparing belief in God to belief in fairy tales.
He shows little interest in the many-worlds interpretation of quantum mechanics, finding it far-fetched.
Hofstadter's current project is a second book on his art, titled 'My Wild Grace Chase.'
His vision of utopia involves global cooperation and peace, lamenting current conflicts like the situation in Gaza and expressing hope for a more compassionate world.
The article discusses how scaling large language models (LLMs) doesn't solve robotics problems because robots need to understand physics, not just language.
V-JEPA 2 is introduced as a solution, trained on 1 million hours of YouTube videos to predict the next moment in reality, not just the next word.
The model uses a ViT-g encoder with 1 billion parameters to understand physical situations and a predictor to fill in masked video segments.
V-JEPA 2-AC extends this by adding a transformer to predict outcomes of actions, trained on just 62 hours of raw robot footage.
The model demonstrates zero-shot generalization, working in new environments with different objects and lighting, achieving high success rates in tasks like reaching and grasping.
Planning with V-JEPA 2-AC is significantly faster than with diffusion models (16 seconds vs. 4 minutes per action).
The model also performs well on video question answering when aligned with a language model, challenging the notion that language supervision is necessary for understanding the world.
Limitations include sensitivity to camera positioning, long-horizon planning drift, and the need for visual goals instead of language instructions.
Future possibilities include world models that rival text models in real-world grounding and robots that understand physics as well as ChatGPT understands language.
New entry-level UK jobs have dropped by 32% since ChatGPT's launch in November 2022.
Businesses are using AI to improve efficiency and reduce staff numbers, with some predicting deeper job cuts.
AI could eliminate half of all entry-level office jobs in the next five years, increasing unemployment by 10-20%.
Graduates face the toughest job market since 2018, with a 33% drop in advertised roles compared to last year.
Companies like Klarna and IBM are using AI for customer service and HR tasks, reducing human roles but increasing demand for programmers and salespeople.
60% of jobs in advanced economies are exposed to AI, with half potentially negatively affected, though new roles may be created.
Workers with AI skills earned 56% more than those without in 2023, up from 25% in 2022.
Employers' skill demands are changing 66% faster in AI-exposed roles, making it harder for workers to keep up.
The UK technology secretary urges workers and businesses to adopt AI now to avoid being left behind.