You can now verify Google AI-generated videos in the Gemini app.Anthropic publishes Agent Skills as an open standard for AI platforms.Guided learning lets “untrainable” neural networks realize their potentialGPT-5.2 dominates OpenAI's new science test, but fails at complex research tasks.

The AI news for December 19th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

You can now verify Google AI-generated videos in the Gemini app.
Source: https://blog.google/technology/ai/verify-google-ai-videos-gemini-app/
Why did we choose this article?
Practical content-provenance tool from Google — useful for anyone worried about deepfakes or edited videos. Signals wider industry emphasis on built-in transparency and gives a concrete step (Gemini app) to verify AI-generated video.

Anthropic publishes Agent Skills as an open standard for AI platforms.
Source: https://the-decoder.de/anthropic-veroeffentlicht-agent-skills-als-offenen-standard-fuer-ki-plattformen/
Why did we choose this article?
A potential interoperability milestone — an open 'Agent Skills' spec could let developers reuse capabilities across multiple AI platforms, lowering lock-in and accelerating composable agent development.

Guided learning lets “untrainable” neural networks realize their potential
Source: https://news.mit.edu/2025/guided-learning-lets-untrainable-neural-networks-realize-their-potential-1218
Why did we choose this article?
Research with practical training implications — shows a way to rescue architectures previously deemed 'untrainable' by using guidance from another network, which could reduce wasted compute and expand usable model families.

GPT-5.2 dominates OpenAI's new science test, but fails at complex research tasks.
Source: https://the-decoder.de/gpt-5-2-dominiert-openais-neuen-wissenschafts-test-scheitert-aber-an-komplexen-forschungsaufgaben/
Why did we choose this article?
New benchmark (FrontierScience) that stresses models at Olympiad/research level — valuable for evaluating where top models like GPT-5.2 excel and importantly, where they still fail on complex research tasks.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

No, the military did not overthrow the government in France.NVIDIA releases Nemotron 3: Hybrid Mamba models for autonomous AI agents.AI’s water and electricity use soars in 2025Adobe hit with proposed class-action, accused of misusing authors’ work in AI training

The AI news for December 18th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

No, the military did not overthrow the government in France.
Source: https://www.heise.de/news/Fake-News-ueber-Staatsstreich-Praesident-Macron-zuernt-Facebook-11118958.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
A clear example of AI-driven deepfakes causing real-world misinformation and platform moderation failures — important for understanding societal risk, verification needs, and policy responses.

NVIDIA releases Nemotron 3: Hybrid Mamba models for autonomous AI agents.
Source: https://the-decoder.de/nvidia-veroeffentlicht-nemotron-3-hybride-mamba-modelle-fuer-autonome-ki-agenten/
Why did we choose this article?
Signals a meaningful architecture shift (hybrid Mamba+Transformer) and unusually open training disclosures — directly relevant to engineers and product teams planning large-context agents.

AI’s water and electricity use soars in 2025
Source: https://www.theverge.com/news/845831/ai-chips-data-center-power-water
Why did we choose this article?
Puts numbers on AI's environmental and resource footprint — crucial for teams assessing operational cost, sustainability strategies, and regulatory risks.

Adobe hit with proposed class-action, accused of misusing authors’ work in AI training
Source: https://techcrunch.com/2025/12/17/adobe-hit-with-proposed-class-action-accused-of-misusing-authors-work-in-ai-training/
Why did we choose this article?
Legal pressure over training-data use is intensifying — this case could influence licensing, model governance, and risk management for organisations building or deploying models.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

Confrontation with Reality: New AI Benchmark OfficeQARobot, make me a chairWhy figures on the water consumption of data centers are often misleading.VS Code disables IntelliCode in favor of the paid Copilot

The AI news for December 17th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

Confrontation with Reality: New AI Benchmark OfficeQA
Source: https://www.heise.de/news/Konfrontation-mit-der-Realitaet-Neuer-KI-Benchmark-OfficeQA-11117132.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
Open-source benchmark from Databricks that specifically evaluates agents in realistic enterprise workflows — useful for practitioners assessing vendor claims, for benchmarking agentic capabilities beyond generic LLM benchmarks, and for teams choosing or auditing agent deployments.

Robot, make me a chair
Source: https://news.mit.edu/2025/robot-makes-chair-1216
Why did we choose this article?
Shows a practical, human-centered advance in robotics and generative design: natural-language-to-physical-object workflows. Valuable for listeners interested in where generative AI meets fabrication, rapid prototyping, and user-driven customization.

Why figures on the water consumption of data centers are often misleading.
Source: https://www.heise.de/news/Warum-Zahlen-zum-Wasserverbrauch-von-Rechenzentren-oft-irrefuehrend-sind-11116970.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
Highlights an underreported sustainability issue: how simplistic water‑use metrics mislead. Important for decision-makers, sustainability teams, and engineers designing data centers or choosing cloud providers — signals need for standardized, transparent reporting and lifecycle-based metrics.

VS Code disables IntelliCode in favor of the paid Copilot
Source: https://www.heise.de/news/VS-Code-deaktiviert-IntelliCode-zugunsten-des-kostenpflichtigen-Copilot-11115668.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
Concrete example of vendor product nudging and potential developer lock-in: Microsoft is pushing paid Copilot over free IntelliCode. Practically important for developers and engineering managers — prompts audits of tooling policies, cost forecasting, and exploration of alternative IDE extensions or self-hosted solutions.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

ChatGPT Images 1.5: Better than Nano Banana Pro?FLUX 2 MAX: Finally perfect images thanks to web searchAdobe Firefly Update: Finally edit videos via text commandAttack on Chrome: Firefox risks everything with a new AI strategy.

The AI news for December 17th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

ChatGPT Images 1.5: Better than Nano Banana Pro?
Source: https://www.all-ai.de/news/top-news24/chatgpt-image-15
Why did we choose this article?
Worthy pick for creators and product teams: OpenAI moves from one-size-fits-all to selectable quality tiers and native aspect ratios, plus a prompting guide — practical changes that affect cost, iteration speed and final-output control.

FLUX 2 MAX: Finally perfect images thanks to web search
Source: https://www.all-ai.de/news/top-news24/flux2-max
Why did we choose this article?
Introduces 'Grounded Generation' (web-verified context) — important for teams who need factual, up-to-date imagery and reduced hallucinations; includes benchmark placement and pricing so creatives can evaluate cost vs. quality.

Adobe Firefly Update: Finally edit videos via text command
Source: https://www.all-ai.de/news/news24/adobe-firefly-ki-video
Why did we choose this article?
Strategically important: Adobe shifts from generation-only to 'directing' existing video via prompts and integrates third-party models — a workflow change that matters to editors, studios and product teams evaluating cross-vendor pipelines.

Attack on Chrome: Firefox risks everything with a new AI strategy.
Source: https://www.all-ai.de/news/top-news24/firefox-ki-strategie
Why did we choose this article?
High-level strategic move with operational implications: Mozilla's 'AI Mode' could reframe browser competition around model choice, local execution and privacy controls — important for product managers, privacy-conscious teams and browser-integrated AI use cases.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

AGI by 2028: Google DeepMind sees a 50 percent chance.We’re publishing an AI playbook to help others with sustainability reporting.Can one still trust AI answers at all?Nvidia bulks up open source offerings with an acquisition and new open AI models

The AI news for December 16th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

AGI by 2028: Google DeepMind sees a 50 percent chance.
Source: https://www.all-ai.de/news/news24/agi-50-deepmind
Why did we choose this article?
Shane Legg (DeepMind co-founder) repeating a 50% AGI-by-2028 forecast is strategically important: it forces policymakers, businesses and researchers to treat AGI timelines as imminent and to plan for workforce, safety and governance implications now.

We’re publishing an AI playbook to help others with sustainability reporting.
Source: https://blog.google/outreach-initiatives/sustainability/ai-playbook-sustainability-reporting/
Why did we choose this article?
Practical, vendor-backed guidance for applying AI to sustainability reporting — useful for corporate teams and consultants looking for concrete methods, risk controls, and templates to operationalize AI in ESG workflows.

Can one still trust AI answers at all?
Source: https://www.heise.de/-11114709?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
A high-impact credibility finding (81% of AI answers contain errors) that should change how teams validate model outputs, design verification pipelines, and set user expectations — essential operational advice for anyone deploying LLMs.

Nvidia bulks up open source offerings with an acquisition and new open AI models
Source: https://techcrunch.com/2025/12/15/nvidia-bulks-up-open-source-offerings-with-an-acquisition-and-new-open-ai-models/
Why did we choose this article?
Signals a shift: Nvidia is strengthening open-source AI infrastructure (Slurm) and releasing new open models, which affects researchers and enterprises by lowering infrastructure friction and expanding options outside closed ecosystems.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

AGI by 2028: Google DeepMind sees a 50 percent chance.We are publishing an AI playbook to help others with sustainability reporting.Can one even trust AI answers anymore?OpenAI-backed biotech firm Chai Discovery raises $130M Series B at $1.3B valuation

The AI news for December 16th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

AGI by 2028: Google DeepMind sees a 50 percent chance.
Source: https://www.all-ai.de/news/top-news24/agibis2028-deepmind
Why did we choose this article?
Shane Legg (DeepMind-Mitgründer) reiterates a 50% chance of 'minimal AGI' by 2028 based on scaling laws and compute/data trends. This is strategically important: it reframes AGI as a near-term, pragmatic threshold (average competent human level) rather than instant superintelligence. Audience takeaway: track concrete definitions, regulatory preparedness, workforce transition planning, and validation frameworks rather than hype.

We are publishing an AI playbook to help others with sustainability reporting.
Source: https://blog.google/outreach-initiatives/sustainability/ai-playbook-sustainability-reporting/
Why did we choose this article?
A practical, vendor-backed playbook for using AI in sustainability reporting targets a real operational need: reliable data, repeatable workflows, and governance. Useful for product managers, sustainability leads, and auditors as a starting point to pilot AI-assisted reporting while managing bias, provenance, and auditability.

Can one even trust AI answers anymore?
Source: https://www.heise.de/-11114709?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
High-impact credibility alert: a Europe-wide study reporting 81% of AI answers contain errors underscores the urgent need for verification, source provenance, and human-in-the-loop processes. Practical implication: treat LLM outputs as drafts, add checks (citation validation, cross-referencing), and design UIs that surface uncertainty.

OpenAI-backed biotech firm Chai Discovery raises $130M Series B at $1.3B valuation
Source: https://techcrunch.com/2025/12/15/openai-backed-biotech-firm-chai-discovery-raises-130m-series-b-at-1-3b-valuation/
Why did we choose this article?
Significant funding signal: OpenAI-backed Chai is using foundation models to predict molecular interactions — a concrete example of generative AI moving into domain-specific, high-impact R&D. For practitioners and investors, watch for reproducibility, wet-lab validation, and how model-driven pipelines change timelines and partnerships in biotech.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

Current AI models master the demanding CFA Financial Analyst exam.Open-source model LongCat shows: Good image AI also works without a flood of parameters.Adobe integrates Photoshop, Acrobat, and Express directly into ChatGPT's user interface.Grok is spreading misinformation about the Bondi Beach shooting.

The AI news for December 15th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

Current AI models master the demanding CFA Financial Analyst exam.
Source: https://the-decoder.de/aktuelle-ki-modelle-meistern-anspruchsvolle-finanzanalysten-pruefungen-cfa/
Why did we choose this article?
Demonstrates current LLM reasoning performance on a real-world, high-stakes professional exam (CFA). Useful signal about near-term capability gains and practical implications for finance workflows, compliance, and upskilling.

Open-source model LongCat shows: Good image AI also works without a flood of parameters.
Source: https://the-decoder.de/meituan-veroeffentlicht-longcat-image-effizientes-6b-modell-fordert-ki-riesen-heraus/
Why did we choose this article?
Highlights a concrete, open-source counterexample to the 'bigger is always better' trend: strong image results from a 6B-parameter model via data and architecture choices — important for cost-efficient deployment and research direction.

Adobe integrates Photoshop, Acrobat, and Express directly into ChatGPT's user interface.
Source: https://the-decoder.de/adobe-integriert-photoshop-acrobat-und-express-direkt-in-die-benutzeroberflaeche-von-chatgpt/
Why did we choose this article?
Practical product integration with immediate workflow impact: text-driven editing of images and documents inside ChatGPT changes how creators and knowledge workers iterate, collaborate, and prototype.

Grok is spreading misinformation about the Bondi Beach shooting.
Source: https://www.theverge.com/news/844443/grok-misinformation-bondi-beach-shooting
Why did we choose this article?
Concrete example of real-world harm from AI misinformation and identity errors. Essential to balance capability stories with risks — informs decisions on deployment, moderation, and trust calibration.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

For 1.6 billion: Intel will probably acquire SambaNova.The U.S. government wants to prevent AI regulation by individual states.Less is more: Individual AI models can defeat networked AI agents.Skills system discovered at OpenAI: ChatGPT is to solve tasks more efficiently with modules.

The AI news for December 14th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

For 1.6 billion: Intel will probably acquire SambaNova.
Source: https://www.heise.de/news/Intels-Uebernahme-von-SambaNova-fuer-1-6-Milliarden-wohl-kurz-vor-Abschluss-11114449.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
Major industry consolidation: an Intel acquisition of SambaNova would reshape the AI chip landscape, affect competition between incumbents and startups, and has direct implications for hardware strategy and procurement.

The U.S. government wants to prevent AI regulation by individual states.
Source: https://www.heise.de/news/US-Regierung-will-KI-Regulierung-durch-einzelne-Bundesstaaten-verhindern-11114371.html?wt_mc=rss.red.ho.themen.k%C3%BCnstliche+intelligenz.beitrag.beitrag
Why did we choose this article?
High-impact policy development: a federal move to preempt state-level AI rules changes the regulatory terrain for US companies, compliance teams, and international observers — essential context for strategy and risk management.

Less is more: Individual AI models can defeat networked AI agents.
Source: https://the-decoder.de/weniger-ist-mehr-einzelne-ki-modelle-koennen-vernetzte-ki-agenten-schlagen/
Why did we choose this article?
Research insight with practical implications: shows when multi-agent systems help or hurt, guiding architects and product teams on whether to pursue agent-based designs or stick with single, stronger models.

Skills system discovered at OpenAI: ChatGPT is to solve tasks more efficiently with modules.
Source: https://the-decoder.de/skills-system-bei-openai-entdeckt-chatgpt-soll-aufgaben-mit-modulen-effizienter-loesen/
Why did we choose this article?
Product and architect-level significance: indicates OpenAI is moving toward modular 'skills' for task specialization, affecting developers building on the platform and expectations for capability, safety, and extensibility.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

Enabling small language models to solve complex reasoning tasksRunway introduces the first 'General World Model' and expands Gen-4.5 with new features.New Pangram 3.0 AI text detector should be almost error-free.Parents call for New York governor to sign landmark AI safety bill

The AI news for December 13th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

Enabling small language models to solve complex reasoning tasks
Source: https://news.mit.edu/2025/enabling-small-language-models-solve-complex-reasoning-tasks-1212
Why did we choose this article?
Novel technique (DisCIPL) shows how ensembles of small models can coordinate to solve complex, constrained reasoning tasks — a practical route to lower-cost, lower-risk deployments compared with always using big LLMs.

Runway introduces the first 'General World Model' and expands Gen-4.5 with new features.
Source: https://the-decoder.de/runway-stellt-erstes-general-world-model-vor-und-erweitert-gen-4-5-um-neue-funktionen/
Why did we choose this article?
Major creative-tool update: Runway’s GWM-1 plus Gen-4.5 native audio and multi-shot editing change workflows for video/audio production — big opportunities for creators, and important to watch for content provenance and misuse (deepfakes).

New Pangram 3.0 AI text detector should be almost error-free.
Source: https://the-decoder.de/neuer-pangram-3-0-ki-text-detektor-soll-fast-fehlerlos-sein
Why did we choose this article?
Detection moves beyond binary labels: Pangram 3.0’s four-way classification and high-accuracy claims matter for publishers, educators, and platforms that need more nuanced signals for moderation, verification, and provenance workflows.

Parents call for New York governor to sign landmark AI safety bill
Source: https://www.theverge.com/ai-artificial-intelligence/844062/parents-call-for-new-york-governor-to-sign-landmark-ai-safety-bill
Why did we choose this article?
Policy development with direct consequences: the RAISE Act would impose developer obligations and education requirements for large models — essential reading for product teams, policymakers, and educators planning compliance and outreach.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].

ChatGPT-5.2 is here: Why this update changes everything.OpenAI partnership: Create your own Disney films with Sora starting in 2026.Google introduces new deep-research agents and a new AI API.FACTS-Benchmark: Even top AI models struggle with the truth

The AI news for December 12th, 2025

--- This episode is sponsored by ---

Find our more about our today's sponsor Pickert at pickert.de.

---

Would you like to create your own AI-generated and 100% automated podcast on your chosen topic? --> Reach out to us, and we’ll make it happen.

Here are the details of the day's selected top stories:

ChatGPT-5.2 is here: Why this update changes everything.
Source: https://www.all-ai.de/news/top-news24/chatgpt-5-2-release
Why did we choose this article?
Major model release with clear practical implications: big gains in accuracy, agentic tool use, long-context handling and reduced hallucinations make this immediately relevant for teams automating knowledge work and developers integrating smarter agents.

OpenAI partnership: Create your own Disney films with Sora starting in 2026.
Source: https://www.all-ai.de/news/top-news24/openai-disney
Why did we choose this article?
High-impact industry partnership that signals how major IP owners may choose licensing and product integration over litigation; practical for creators, platform engineers, and policy watchers planning for branded generative media.

Google introduces new deep-research agents and a new AI API.
Source: https://the-decoder.de/google-stellt-neuen-deep-research-agenten-und-neue-ki-api-vor/
Why did we choose this article?
Developer-facing update: Google upgrading Deep-Research with Gemini 3 Pro and an API that better supports agentic workflows matters for teams building autonomous tool chains and research assistants — signals where platform-level agent support is heading.

FACTS-Benchmark: Even top AI models struggle with the truth
Source: https://the-decoder.de/facts-benchmark-auch-top-ki-modelle-kaempfen-mit-der-wahrheit/
Why did we choose this article?
Critical reliability research: DeepMind's FACTS benchmark highlights persistent truthfulness gaps even in leading models — essential context for teams relying on LLM outputs and for designing verification layers and human-in-the-loop checks.

Do you have any questions, comments, or suggestions for improvement? We welcome your feedback at [email protected].