Meta Launches Muse Glimmer 30B and $1B Fund Amid Leaked Employee-Tracking and Performance Scandals
Meta Platforms has dramatically expanded its open-weight AI strategy, launching a new model, a $1 billion community fund, and a sweeping 6,500-word manifesto by CEO Mark Zuckerberg. However, leaked internal recordings and executive statements have exposed a sharp divergence between Meta’s public-facing utopian rhetoric and its internal operational realities, which are marked by employee surveillance, security failures, and sluggish developmental progress.
Public Expansion: Muse Glimmer 30B and Zuckerberg's Manifesto
On August 10, 2026, Meta officially released the weights for Muse Glimmer, a 30-billion-parameter dense model optimized for local agentic workflows, multi-step reasoning, and coding on consumer hardware (such as Macs or PCs with a single GPU). Released under the permissive Apache 2.0 license, Muse Glimmer is positioned as a direct counterweight to the closed-source API ecosystems of OpenAI and Anthropic.
Alongside the model, Zuckerberg published a lengthy manifesto titled "The Future is for Everyone: The Path to a Positive AI Future." In it, he outlines a vision of "personal superintelligence" where every individual possesses a highly capable personal agent. Zuckerberg also took aim at closed-source competitors and criticized regulatory efforts to limit model distillation, writing:
"Some have tried to frame distillation as harmful, but I think it is important to protect the principle that you can learn from anything you can observe. This is how the world works, and the US will not be able to lead if we restrict ourselves on this front."
To resolve the public relations crisis surrounding the massive energy and water consumption of its infrastructure, Meta simultaneously announced the $1 billion "Future Is For Everyone Fund." The fund will invest in schools, public services, and infrastructure in towns hosting Meta's sprawling data centers, such as Richland Parish, Louisiana.
Internal Reality: The "Model Capability Initiative" Surveillance Scandal
While Zuckerberg's public manifesto champions "individual empowerment as the source of prosperity," leaked recordings of internal town halls tell a drastically different story about Meta's workplace practices.
According to a series of leaked recordings, in April 2026, Meta quietly installed tracking software on its U.S. employees' company laptops through a program called the Model Capability Initiative (MCI). The software logged:
- Every keystroke, mouse movement, and click.
- Periodic screen captures and snapshots of content.
- Activity across work-related applications and websites.
The harvested telemetry was fed directly into Meta's AI models as training data, essentially using its own engineers' coding behaviors as an unwitting source of raw training material. The initiative triggered an immediate employee revolt, with more than 1,600 employees signing a petition demanding its termination and staff in the United Kingdom launching a unionization drive in response.
The program ultimately collapsed due to a severe internal security failure. On June 22, 2026, Meta was forced to pause the MCI program after an internal security review discovered that the harvested employee data—including private conversations, prompts, and performance-related records—had been left exposed and accessible across roughly 45,000 internal database tables.
Executive Backlash and Sluggish Performance
The friction between leadership and staff has been further exacerbated by comments from Meta CTO Andrew Bosworth. During an internal Q&A session, an employee asked if the productivity gains from AI would allow workers to take more time off. Bosworth aggressively rejected the premise, stating:
"I hope that what we do with our extra time is to do even more and cooler stuff for the users who use our products every day... Go to your parents and ask them: hey, like every time I get a chance to talk to my boss, ask me if I can have more days off. Ask your parents what they think of that as a career strategy."
Despite these aggressive demands and invasive tracking efforts, Meta's underlying AI progress has stalled. In a leaked recording of a closed-door town hall on July 2, 2026, Zuckerberg admitted to employees that the company's AI push "hasn't really accelerated in the way that we expected" over the preceding four months.
This internal admission severely undermines Meta’s public-facing narrative of rapid, frictionless superintelligence development, highlighting that the company's aggressive, surveillance-driven AI training push has failed to yield the commercial and technical breakthroughs promised to investors.