The AI & Tech Society by Danar

Claude Code: How Anthropic is using Claude Code

Boris Cherny, Head of Claude Code:

Dario Amodei, CEO:

Jack Clark, Co-founder:

1. Invest in CLAUDE.md documentation — Configuration files Claude reads at startup
2. Classify tasks: async vs synchronous — Know what to supervise vs delegate
3. Create self-sufficient verification loops — Tests before code, auto-run builds/lints
4. Start from clean git state — Checkpoint commits enable safe experimentation
5. Use MCP servers for sensitive data — Better logging and access control
6. Build multi-instance parallel workflows — Multiple Claude instances across repos
7. Use screenshots and multimodal input — Figma, dashboards, UI images
8. Prompt for simplicity — Interrupt and ask "Try something simpler"

The AI PM Cert visit: https://aipmcert.com/

Hosted on Acast. See acast.com/privacy for more information.

Episode: What 81,000 People Want From AI: The Most Human AI Report So Far

Study: Anthropic Global AI Survey (December 2025)

• 80,508 Claude users interviewed
• 159 countries
• 70 languages
• AI-conducted open-ended conversations

CategoryPercentageProfessional Excellence18.8%Personal Transformation13.7%Life Management13.5%Time Freedom11.1%Financial Independence9.7%

Key insight: Productivity is often the surface story. When asked what productivity enables, people reveal deeper wants: family time, mental health, meaningful work, paths out of precarity.

Hosted on Acast. See acast.com/privacy for more information.

Pentagon Position:

• Requires "all lawful use" provisions from AI vendors
• Wants flexibility for future applications
• Focused on Golden Dome, drone swarms, autonomous systems

Anthropic Position:

• Two non-negotiables: no mass surveillance of Americans, no fully autonomous weapons
• Will not sign contracts creating legal pathways to prohibited uses
• Challenging supply chain risk designation in court

OpenAI Position:

• Explicit contractual prohibitions on mass surveillance, autonomous weapons, high-stakes automated decisions
• Cloud-only deployments with OpenAI personnel in loop
• Maintains control over safety stack

Current Uses:

• Intelligence analysis
• Cyber operations
• Operational planning
• Threat assessment
• Modeling and simulation
• Classified environment support

Hosted on Acast. See acast.com/privacy for more information.

Episode: AI and Jobs in 2026: What Anthropic's Labor Report Really Means for Workers, Policy, and Business

Report: Anthropic Economic Index Labor Market Analysis (March 5, 2026)

No mass displacement yet, but entry is getting harder:

• No systematic increase in unemployment for AI-exposed occupations
• Job-finding rates for workers aged 22-25 in exposed fields: down ~14% vs 2022
• Unemployment rates: flat
• First visible effect: fewer young people getting their first foothold

ComponentWhat It MeasuresTheoretical Capability% of tasks LLMs could theoretically performObserved UsageWhat people actually do with Claude at workObserved ExposureCombined measure weighted toward automated/work-related uses

Why it matters: Labor markets are shaped by adoption, workflow design, regulation, and trust—not just model demos.

Hosted on Acast. See acast.com/privacy for more information.

All three dropped same day:

• OpenAI: GPT-5.3-Codex-Spark (purpose-built for engineering workflows)
• Google: Gemini 3 Deep Think
• Anthropic: Major funding round announcement

Three-way battle for developer mindshare officially a sprint

Framework: Five "Priority Sprint Projects"

Initiatives:

• GenAI.mil for all-classification AI access
• Enterprise agents playbook

Mandate: All military departments must identify 3+ priority AI projects within 30 days

Language:

• "Any lawful use" in procurement
• "Military AI dominance" framing

Action: Cease-and-desist letters (Feb 14)

Target: Seedance 2.0 video generation

Accusation: Generating copyrighted characters (Star Wars, Marvel)

MPA Statement: "Unauthorized use of U.S. copyrighted works on a massive scale"

Implication: AI copyright fight moves from theoretical to legal

Source: UC Berkeley study in Harvard Business Review

Finding: AI users worked faster, took on more tasks, worked longer hours—often without being asked

Implication: AI isn't reducing workload—it's intensifying it

Recommendation: Managers must design for outcomes, not just output

Releases (mid-February):

• DeepSeek V4: 1 trillion parameters, coding-focused
• Alibaba Qwen 3.5
• ByteDance Doubao upgrade

Cost Advantage (RAND): Chinese models run at 1/6 to 1/4 cost of comparable U.S. systems

Market Share: DeepSeek holds ~89% among AI users in China

Announcement (Feb 12): Top developers haven't manually written code since December

Tools:

• Claude Code
• Internal system "Honk"

Shift: Engineers are now "full-time AI orchestrators"

Implication: Future of engineering is operational, not hypothetical

1. Commercialization-safety tension is real — Ads + safety team dissolution not coincidental
2. Brand positioning matters — 11% user bump from values messaging
3. Coding model wars intensifying — Three releases same day
4. Government AI accelerating — 30-day Pentagon mandate
5. Copyright enforcement getting real — Disney vs. ByteDance
6. AI may increase workload — Design for outcomes, prevent burnout

OpenAI, Anthropic, Google, Disney, Paramount, ByteDance, Spotify, DeepSeek, Alibaba, Motion Picture Association, Department of Defense

• Sam Altman (OpenAI CEO)
• Joshua Achiam (OpenAI, now "chief futurist")

• UC Berkeley/HBR: AI and workload intensification
• BNP Paribas: Super Bowl ad effectiveness
• RAND: Chinese AI cost analysis

Hosted on Acast. See acast.com/privacy for more information.

Today we're doing something different. Instead of covering the news cycle, we're going deep on the three people who will likely shape how AGI arrives: Sam Altman of OpenAI, Dario Amodei of Anthropic, and Demis Hassabis of Google DeepMind.

Each has a distinct philosophy about how to build transformative AI, what the risks are, and what happens to society when we get there. Understanding these differences isn't academic. These philosophies are determining the products we use, the policies being debated, and potentially the trajectory of human civilization.

Hosted on Acast. See acast.com/privacy for more information.

1. Model race is now platform race (Cowork vs Frontier)
2. $650B Big Tech capex is the new reality
3. Professional software under genuine threat
4. Hardware competition intensifying (AMD, Broadcom)
5. Regulatory complexity growing (federal vs state)
6. AI adoption mainstream but returns concentrated
7. Super Bowl ads signal consumer battleground

Anthropic, OpenAI, Alphabet/Google, Amazon, Meta, Microsoft, NVIDIA, AMD, Broadcom, Thomson Reuters, LegalZoom, HP, Intuit, Oracle, State Farm, Uber, Cisco, BBVA, T-Mobile, Cerebras, Goodfire, Bedrock Robotics, Sana, Perplexity, Boston Dynamics, Caterpillar, Khan Academy

Hosted on Acast. See acast.com/privacy for more information.

ProductCompanyDomainKey FeaturePrismOpenAIScienceGPT-5.2 with 400K token context for researchGOV.UK AssistantAnthropicGovernmentAgentic employment support for UKPersonal IntelligenceGoogleConsumerGmail + Photos integration in AI ModeAI Overviews upgradeGoogleSearchGemini 3 default, follow-up questions

Model: GPT-5.2

• 400,000-token context window (~800 pages)
• Fine-tuned for mathematical and scientific reasoning
• Native LaTeX understanding
• Visual Synthesis for diagrams to code

Pricing:

• Personal: Free (unlimited projects/collaborators)
• Education: Institutional tier (TBD)
• Enterprise: Compliance features (TBD)

Built on: Acquired startup Crixet (LaTeX platform)

Competition:

• Overleaf (LaTeX collaboration)
• Mendeley/Zotero (reference management)
• Google Scholar integration (anticipated)

Name: Transparency and Responsibility for Artificial Intelligence Networks Act

Sponsors: Rep. Madeleine Dean (D-PA), Rep. Nathaniel Moran (R-TX)

Key Provisions:

• Administrative subpoena for training data disclosure
• "Subjective good faith belief" standard for requests
• Non-compliance creates "rebuttable presumption of copying"

Impact: Gives copyright holders discovery rights previously unavailable

ASML Q4 2025:

• Orders: €13.2B ($15.8B) — record quarter
• Analyst forecast: €6.85B (far exceeded)
• Q4 sales: €9.72B
• Full 2025 sales: €32.7B
• Stock surge: ~6%

Intel: Activated ASML EXE:5200 High-NA EUV system

• Reduces manufacturing steps: 40 → 10

Spending Forecasts (Gartner):

• 2026: $2.53 trillion
• 2027: $3.33 trillion

Hosted on Acast. See acast.com/privacy for more information.

Hosted on Acast. See acast.com/privacy for more information.

Kristalina GeorgievaIMF40% of jobs "touched by AI," labor market "tsunami"

Dario AmodeiAnthropicAI will replace all developers in 1 year, Nobel-level science in 2 years

Demis HassabisGoogle DeepMindAGI has 50% chance within decade, current systems "nowhere near" human-level

Satya NadellaMicrosoftEnergy costs, not model quality, will determine AI winners

Donald TrumpU.S. President300GW nuclear capacity target by 2030

Jensen HuangNVIDIARobotics is "once-in-lifetime opportunity" for Europe

Yuval Noah HarariPhilosopherWarning about AI manipulation potential

OpenAI Consumer Device

• Confirmed for H2 2026 launch
• Likely screenless, possibly wearable
• Healthcare applications following Torch Health acquisition

Thinking Machines Crisis

• Three co-founders left: Barret Zoph, Luke Metz, Sam Schoenholz
• All returned to OpenAI
• Murati fired Zoph citing "unethical conduct"
• $12B startup facing existential talent crisis

Lightricks LTX-2 Launch

• Audio-to-video generation with ElevenLabs partnership
• Audio becomes control layer for video generation
• 4K/50fps output capability
• Open-source with complete training code
• API access launched January 27

• Data center power: 55GW current → 84GW by 2028 (Goldman Sachs)
• Trump targeting 300GW new nuclear by 2030
• Nadella: "Tokens are new commodity"
• Energy costs = AI competitive advantage

SourceFindingIMF40% of jobs touched by AIBCG AI Radar50% of CEOs believe jobs depend on AI successMcKinseyTwo-thirds haven't scaled AI enterprise-wideMicrosoft ResearchFinance, legal, software engineering most exposedGartner$1.5 trillion invested in AI in 2025

ProductCompanyCapabilityLTX-2 Audio-to-VideoLightricks + ElevenLabsGenerate video from audio controlChat with PDFAdobe AcrobatPrompt-based PDF editing and summariesReal Talk + VideoMicrosoft CopilotHuman-like conversation, video generationClaude in ChromeAnthropicBrowser extension for in-page AI assistanceAuth0 for AI AgentsAuth0Identity management for AI agentsScroll AIScrollKnowledge base to AI expert conversion

Architecture:

• 14B parameter video stream
• 5B parameter audio stream
• Cross-attention for synchronization
• Single diffusion pass generation

Capabilities:

• 4K/50fps output
• 10-20 second clips
• LoRA fine-tuning in under 1 hour
• Audio-first control paradigm

Open Source:

• Complete training code released
• Architecture documentation
• Community fine-tuning enabled

1. Workforce transformation is urgent (40% jobs affected)
2. Energy strategy is AI strategy (tokens as commodity)
3. AI startup ecosystem is volatile (talent wars)
4. New form factors coming (OpenAI device H2 2026)
5. Audio-video generation is production-ready (LTX-2)
6. ROI pressure is real (2/3 failing to scale)
7. AI safety includes manipulation risks (Harari warning)

• Kristalina Georgieva (IMF Managing Director)
• Dario Amodei (Anthropic CEO)
• Demis Hassabis (Google DeepMind CEO)
• Satya Nadella (Microsoft CEO)
• Jensen Huang (NVIDIA CEO)
• Donald Trump (U.S. President)
• Mira Murati (Thinking Machines CEO, former OpenAI CTO)
• Chris Lehane (OpenAI Chief Global Affairs Officer)
• Yuval Noah Harari (Philosopher)
• Barret Zoph, Luke Metz, Sam Schoenholz (former Thinking Machines)

OpenAI, Anthropic, Google DeepMind, Microsoft, NVIDIA, Meta, Thinking Machines, Lightricks, ElevenLabs, Adobe, Auth0, Scroll AI, DeepSeek, Moonshot AI, Runway, Pika

Hosted on Acast. See acast.com/privacy for more information.