Current AI Risk Assessment
Chance of AI Control Loss
Estimated Date of Control Loss
AGI Development Metrics
AGI Progress
Estimated Date of AGI
Risk Trend Over Time
Latest AI News (Last 3 Days)
Anthropic Releases Claude Sonnet 4.6 with Enhanced Coding and 1M Token Context Window
Anthropic has launched Sonnet 4.6, featuring significant improvements in coding, instruction-following, and computer use capabilities, along with a doubled context window of 1 million tokens. The model achieves strong benchmark results including a 60.4% score on ARC-AGI-2, positioning it above most comparable models though still trailing top-tier systems like Opus 4.6 and Gemini 3 Deep Think. This release maintains Anthropic's four-month update cycle and will serve as the default model for Free and Pro users.
Skynet Chance (+0.02%): Improved instruction-following and autonomous computer use capabilities increase potential for more independent AI systems, though the model remains behind the most advanced frontier systems. The incremental nature and continued human oversight mechanisms suggest modest risk elevation.
Skynet Date (+0 days): The sustained four-month release cycle and competitive benchmark improvements demonstrate consistent capability acceleration across the industry. However, the model's position below top-tier systems suggests this represents expected progress rather than breakthrough acceleration.
AGI Progress (+0.02%): The 60.4% ARC-AGI-2 score represents meaningful progress on benchmarks specifically designed to measure human-like general intelligence, alongside substantial improvements in coding and autonomous computer use. The 1 million token context window enables more complex reasoning over larger information sets, advancing toward AGI-relevant capabilities.
AGI Date (+0 days): Anthropic's consistent four-month release cycle with measurable capability gains demonstrates sustained momentum in the industry, accelerating the timeline toward AGI. The fact that mid-tier models are now achieving 60%+ scores on human intelligence benchmarks suggests faster-than-expected progress across the capability spectrum.
Ricursive Intelligence Raises $335M to Build AI-Powered Chip Design Platform
Ricursive Intelligence, founded by former Google Brain and Anthropic engineers Anna Goldie and Azalia Mirhoseini, raised $335 million at a $4 billion valuation to develop AI tools that automate chip design. Their platform, based on their acclaimed Alpha Chip work at Google, uses reinforcement learning to generate chip layouts in hours instead of years, learning and improving across multiple designs. The company aims to accelerate AI advancement by enabling faster co-evolution of AI models and the chips that power them, potentially achieving 10x efficiency improvements.
Skynet Chance (+0.04%): The capability for AI to design its own hardware creates a potential recursive self-improvement loop, reducing human oversight in critical infrastructure design. This increases autonomy and capability scaling, though the founders emphasize efficiency benefits and the technology remains in early commercial stages.
Skynet Date (-1 days): By dramatically accelerating chip design cycles and enabling faster co-evolution of AI models with their underlying hardware, this technology could significantly speed up AI capability advancement. The founders explicitly state this will allow "AI to grow smarter faster," directly accelerating the timeline for advanced AI systems.
AGI Progress (+0.04%): This represents a meaningful advancement toward AGI by addressing a key bottleneck: hardware design speed. The ability to rapidly iterate on specialized AI chips and enable faster co-evolution of models and hardware directly supports the scaling and optimization required for AGI development.
AGI Date (-1 days): The platform substantially accelerates chip development from years to hours and enables rapid hardware-software co-optimization, removing a major constraint on AI advancement pace. The founders explicitly position this as enabling faster AI evolution, with potential 10x efficiency improvements that could dramatically accelerate AGI timelines.
U.S. Universities See CS Enrollment Drop as Students Shift to AI-Specific Programs
Computer science enrollment at UC campuses dropped 6% this fall, with the exception of UC San Diego, which launched a dedicated AI major. While U.S. universities scramble to launch AI-specific programs, Chinese universities have already made AI literacy mandatory and integrated it across curricula, with nearly 60% of students using AI tools daily. American institutions face faculty resistance and are racing to create AI-focused degrees as students increasingly choose specialized AI programs over traditional CS majors.
Skynet Chance (-0.03%): Increased AI literacy and education across broader student populations could lead to more informed development practices and awareness of risks, though it also accelerates the number of people capable of building advanced AI systems. The net effect is slightly positive for safety as understanding risks is the first step toward mitigation.
Skynet Date (-1 days): The massive educational shift toward AI, particularly China's aggressive integration of AI literacy across institutions, will significantly accelerate the development of AI capabilities by producing more AI-trained talent entering the workforce. This educational arms race, especially with 60% of Chinese students already using AI tools daily, compresses the timeline for advanced AI development.
AGI Progress (+0.03%): The systematic integration of AI education at scale, particularly in China where it's now mandatory at top institutions, represents a fundamental shift in human capital development that will accelerate AGI research. More AI-literate graduates entering the field with specialized training creates a stronger talent pipeline for AGI development than traditional CS programs.
AGI Date (-1 days): The rapid expansion of AI-specific degree programs and mandatory AI coursework, especially China's aggressive approach with nearly 60% daily AI tool usage among students, will dramatically accelerate the pace of AGI development by creating a larger, more specialized workforce. This educational transformation represents a structural acceleration in the AGI timeline as universities shift from debating AI integration to producing thousands of AI-specialized graduates annually.
AI News Calendar
AI Risk Assessment Methodology
Our risk assessment methodology leverages a sophisticated analysis framework to evaluate AI development and its potential implications:
Data Collection
We continuously monitor and aggregate AI news from leading research institutions, tech companies, and policy organizations worldwide. Our system analyzes hundreds of developments daily across multiple languages and sources.
Impact Analysis
Each news item undergoes rigorous assessment through:
- Technical Evaluation: Analysis of computational advancements, algorithmic breakthroughs, and capability improvements
- Safety Research: Progress in alignment, interpretability, and containment mechanisms
- Governance Factors: Regulatory developments, industry standards, and institutional safeguards
Indicator Calculation
Our indicators are updated using a Bayesian probabilistic model that:
- Assigns weighted impact scores to each analyzed development
- Calculates cumulative effects on control loss probability and AGI timelines
- Accounts for interdependencies between different technological trajectories
- Maintains historical trends to identify acceleration or deceleration patterns
This methodology enables data-driven forecasting while acknowledging the inherent uncertainties in predicting transformative technological change.