President @drpezeshkian shared the signed Iran–US agreement, stating that it represents both a historic milestone and a testament to Iran's strength.

Not a loser, I'm stating the facts.

They’re only stating the truth.

I agree. Narratives can mess with our heads. But Warsh stating that he was "really open-mined", then reiterating by saying he was "pretty open-minded" must have been a blow. I mean, this is one of the most conservative heads in the country. Look at the neighbourhood he lives in. Very disturbing, isn't it.

Majority Leader Mahama Ayariga calls on Ghanaians to hold the NPP accountable, stating that if the party cannot produce former Finance Minister Ken Ofori-Atta, they should be denied a return to power. #3NewsGH #TV3GH

He is not a good man for stating the truth? y'all are just dumb, if a woman don't fear her father who is she gonna fear but u are too dvll to reason what happened to her being neutral

Yeah, that's because what the Dr. said. If memory serves me correct, it was the Dr. that worked on Charlie at the hospital who suggested that the reason Charlie wasn't decapitated by the 30.06 by stating Charlie was probably shot by a "frangible round."

I know. I’m just stating a fact.

So, you made @joshsimonsmp stand doen and now you are openly stating he wasn't up for the job. Wow. He must hate you. By the way #NeverBurnham #IStandWithStarmer #Traitor

Where is your proof? You got an answer for everything. Just talking out if your ass. Everyone would get emergency medicaid if what youre stating is true. Its not. Try using your critical thinking skills, Angie.

Iranian Vice President Mohammad Reza Aref is claiming victory in the US-Iran war, stating: “Control over the Strait of Hormuz and its management were among the achievements of the Ramadan War.” He added: “From now on, the management of the Strait of Hormuz will be under Iran’s responsibility, and we will collect fees for the services we provide to ships.”

YSRCP spokesperson Are Syamala came down heavily on Home Minister @Anitha_TDP, stating that a person entrusted with maintaining law and order should focus on protecting citizens rather than indulging in abusive language and personal attacks against political opponents. She said that despite holding a key constitutional position, Anitha has repeatedly crossed the limits of political decency and has been attempting to divert public attention from the growing law and order crisis in the state. Syamala alleged that Andhra Pradesh has slipped into a state of fear and insecurity, citing the mysterious disappearance of Gade Sai Krishna from Vijayawada, the prolonged missing case of a young girl in Kakinada, and the controversy surrounding TDP MLA Chintamaneni Prabhakar's alleged actions in Denduluru. She said that while crimes, missing person cases and public safety concerns continue to rise, the Home Minister remains preoccupied with political attacks and publicity. Syamala demanded that Anitha stop resorting to diversionary politics and instead concentrate on restoring law and order, ensuring justice for victims and safeguarding the lives and security of the people of Andhra Pradesh.

I was fine with him stating his cause the first time and I had somewhat agreed. But once he said it for the 100th time it became annoying and disrespectful. The Martin Kabrhel of poker commentating

No, it's not ok, you extremist fool. It's a mobile prison, designed to put women in subservient roles. Or, if the need arrives, smuggle terrorists posing as women. Sure, women can be indoctrinated, or beaten, into stating it's their choice. But it only serves to enslave them.

Thorium Labs AI Daily Digest — June 18, 2026 Good morning, guys. Welcome to my AI Daily Digest, a roundup of the most critical developments in AI, AI Industry Signals, agentic AI, local/self-hosted frontier models, and broader ecosystem signals from the past 24 hours. I focus on the acceleration pushing the frontier forward. INFLUENCER SPOTLIGHTS GREG ISENBERG (@gregisenberg)– Startup ideas, business angles on AI • Highlighted resources for building AI-native products via Late Checkout. • Continued signaling around bootstrapped execution in the AI space. • No deeper threads in the window, but consistent focus on practical business models that leverage the current acceleration. • latecheckout.agency/ ALEX FINN (@AlexFinn)– Vibe-coding, home AI labs, agent swarms • Delivered a widely engaged reaction to the GLM-5.2 release, stating “I was wrong. I’ve been saying for months that open source AI models are 6 months behind frontier. They caught up. GLM 5.2 is as good as Opus 4.8. This changes everything.” • Emphasized local execution for sovereignty (“If you run GLM 5.2 locally no government can take it away. You become sovereign”), dramatic cost reduction versus frontier APIs, and the likely impact on regulation and closed labs. • Noted the model’s size requires significant GPUs today but predicted rapid efficiency and hardware gains will bring it to consumer hardware soon: “This is the worst it will ever be… I’d bet ANYTHING on it.” • x.com/AlexFinn/status/206728… MATTHEW BERMAN (@MatthewBerman)– Local hardware, OpenClaw releases, frontier open models, inference speed • Shared a thoughtful essay on AI consciousness by researcher @camhberg. • Reacted to WIRED reporting on regulatory pressure around Anthropic’s potential Fable 5 rerelease: “guess we aren’t getting fable back…” • Teased upcoming content with “something coming soon” and “building up,” likely tied to local model evaluations or OpenClaw-related work. • x.com/MatthewBerman/status/2… • x.com/MatthewBerman/status/2… PETER STEINBERGER (@steipete)– OpenClaw core maintainer, accessibility, technical reality checks No major new posts in the window; follow for ongoing insights on OpenClaw development, local inference accessibility, and technical reality checks. Teknium (@teknium) - Cofounder & Lead Engineer at Nous Research No major new posts in the window; follow for ongoing insights on frontier open models and the Nous Research ecosystem. KEY AGENTIC AI HIGHLIGHTS • GLM-5.2 delivers significant gains in coding and agentic tasks, long-horizon reasoning with a 1M context window, and dual reasoning modes (max vs high efficiency). MIT-licensed open weights enable true self-hosted frontier capabilities. • OpenAI demonstrated meaningful agentic progress in real scientific research: GPT-5.4 reviewed literature, generated hypotheses, designed experiments, analyzed results, and proposed an optimized Chan-Lam coupling reaction. Human chemists validated substantial yield improvements. • The open-source catch-up narrative has shifted sharply. Local/self-hosted execution is now positioned as a sovereignty play rather than a capability compromise. • Community discussion reinforces that hardware curves and model optimizations will continue making these large models runnable on increasingly accessible systems — personal ownership matters. • Ongoing focus on runtime optimizations, RL-from-chat techniques, and agent infrastructure to support reliable long-running autonomous workflows. AI INDUSTRY SIGNALS OPENAI (@OpenAI)– Frontier models, scientific applications • Released LifeSciBench, a new benchmark developed with 173 scientists containing 750 expert tasks across seven real biological research workflows. GPT-Rosalind outperformed GPT-5.5 across these realistic, artifact-heavy, and operationally constrained tasks. • Achieved a notable milestone in medicinal chemistry collaboration with Molecule.one’s Maria AI, moving from literature review through experimental validation with measurable improvements in pharmaceutical reaction yields. • x.com/OpenAI/status/20673469… • x.com/OpenAI/status/20672937… XAI (@xai)– Grok models, agentic tooling • Expanded easy access to Grok Build (their coding agent) through DigitalOcean Marketplace with one-click VMs and Serverless Inference. • Further Bedrock integration for Grok 4.3 noted in documentation updates. • x.com/xai/status/20673510992… ANTHROPIC (@AnthropicAI)– Safety-focused frontier models • WIRED reported that the Trump administration has signaled strict requirements for any rerelease of Fable 5, specifically demanding jailbreak prevention that security experts believe cannot be fully guaranteed. This highlights persistent tension between regulatory demands and technical reality. BROADER INDUSTRY & MACRO NEWS • The GLM-5.2 release (MIT licensed, strong agentic and coding performance, 1M context) represents one of the clearest signals yet that open-weight models have reached effective parity with certain frontier closed systems in key domains. This shifts cost curves, regulatory leverage, and the open vs closed debate. • Enterprise and scientific adoption continues to accelerate, with OpenAI’s chemistry results providing concrete evidence of AI meaningfully compressing the research loop from hypothesis to validated experiment. • Regulatory pressure on model guardrails and jailbreaks is increasing under the current administration, per WIRED reporting, raising questions about the feasibility of “unbreakable” safety constraints. • Hardware and efficiency narratives remain optimistic. Even models that currently demand high-end GPU clusters are expected to reach widely available consumer hardware within years. • Life sciences and pharmaceutical sectors show particularly strong signals of industrial AI integration, supported by new benchmarks like LifeSciBench that better reflect real workflows. • The overarching theme is continued acceleration: capabilities are advancing on both open and closed fronts, self-hosting viability is improving, and personal ownership of models is becoming an increasingly practical hedge against centralized control. Today's episode highlights a pivotal 24-hour window where open-weight models reached striking parity with frontier performance while closed labs demonstrated tangible scientific acceleration. The acceleration continues, with agentic capabilities, self-hosting wins, and real-world research impact pushing the frontier forward. Personal ownership matters more than ever in this environment. Tune in tomorrow at the same time for the next digest. Your feedback is welcome. #artificialintelligence #web3 #blockchain #AgenticAI #AaaS #AgentsasaService #immersiveexperiences #XR #AR #VR #thoriumlabs #buildingbetterworlds

Your US citizenship should be revoked — go back to Israel! Ana Kasparian blasts Israeli-American dual citizens, stating they must "make a decision": You're either Israeli or an American.

The government's seizure of property without conviction is legalized theft, betraying our Founders' sacred trust. Federal statutes, specifically 18 U.S.C. §§ 981–985, permit the seizure of property suspected of being "involved in" or "obtained through" criminal activity, often without charging the owner. 18 U.S.C. § 981(a)(1)(C) states that property "involved in" or "traceable to property involved in" certain offenses is subject to forfeiture. This allows asset confiscation based on mere suspicion, not proven guilt. This practice directly affronts the Fifth Amendment: "No person shall be... deprived of life, liberty, or property, without due process of law." Our Constitution demands due process, not state-sanctioned plunder. James Madison, in Federalist Paper No. 47, warned against the concentration of power, stating, "The accumulation of all powers... in the same hands... may justly be pronounced the very definition of tyranny." Unchecked forfeiture power exemplifies this tyranny. Such actions are utterly repugnant to the spirit of liberty. A government that may seize a citizen's assets without a conviction has strayed far from its constitutional moorings, making every individual's property precarious. To restore justice, 18 U.S.C. §§ 981–985 must be repealed. Furthermore, the FAIR Act should pass, establishing a clear and convincing evidence standard and increasing transparency. This is not reform, but restoration. #ReclaimSovereignty #FoundersVoice

No. This is a very simple effect that has a bunch of text stating a bunch of restrictions to when you can use it. This isn't overtuned.

calm down dude do not come for me. just stating the possible scenario. The last time i checked you did not buy data for me so next time you cannot order me to stop commenting.

folks will bemoan “woke 1.0” and support a candidate stating insane things like this that 98% of the country disagree with (Also this is not what double jeopardy even means)