From prompts to ongoing work. The September 30 AI briefing, designed for reading: five stories, quick takeaways, and all six sources.
<!doctype html>
<html lang="en"><head><meta charset="utf-8"><meta name="viewport" content="width=device-width, initial-scale=1"><meta name="color-scheme" content="light"><meta name="description" content="Five AI developments: cheaper models, ongoing agents, shared workspaces, small-business operations, and a voluntary safety accord.">
<style>
:root{color-scheme:light;--ink:#23251f;--muted:#696d61;--line:#dce0d6;--paper:#f6f7f2;--accent:#e75b38;--soft:#eef1e7}*{box-sizing:border-box}html{scroll-behavior:smooth}body{margin:0;background:var(--paper);color:var(--ink);font-family:Arial,Helvetica,sans-serif;-webkit-font-smoothing:antialiased}.page{max-width:1100px;margin:auto;padding:0 48px}a{color:inherit;text-decoration:underline;text-decoration-color:#e75b3880;text-underline-offset:4px}a:hover{color:#ab351a}a:focus-visible{outline:2px solid var(--accent);outline-offset:4px}.masthead{padding:29px 0 22px;display:flex;justify-content:space-between;align-items:center;gap:20px;border-bottom:2px solid var(--ink)}.brand{font-size:24px;font-weight:800;letter-spacing:-1.1px}.star{color:var(--accent);font-size:30px;vertical-align:middle;margin-left:4px}.masthead-right{text-align:right;font-size:10px;font-weight:700;line-height:1.6;letter-spacing:1.4px;text-transform:uppercase}.masthead-right span{display:block;color:var(--muted);font-weight:400;letter-spacing:.8px}.hero{padding:37px 0 30px}.eyebrow{font-size:10px;letter-spacing:2px;text-transform:uppercase;font-weight:700;display:flex;gap:12px;align-items:center}.dot{width:7px;height:7px;background:var(--accent);border-radius:50%;display:inline-block}.hero h1{font-size:clamp(42px,6.8vw,78px);line-height:1.02;letter-spacing:-3.8px;font-weight:500;font-family:Georgia,'Times New Roman',serif;margin:23px 0 25px;max-width:900px}.hero h1 em{font-style:italic;color:#63754b}.hero-grid{display:grid;grid-template-columns:1fr 200px;gap:48px;align-items:start}.hero-lead{font-size:16px;line-height:1.75;margin:0;max-width:710px;color:#4c5147}.issue-note{border-left:1px solid var(--line);padding-left:22px;font-size:11px;line-height:1.7;color:var(--muted)}.issue-note strong{display:block;font-size:11px;color:var(--ink);margin-bottom:6px;text-transform:uppercase;letter-spacing:1px}.issue-note span{display:block}.section-bar{display:flex;align-items:center;justify-content:space-between;gap:16px;margin-bottom:17px;border-top:1px solid var(--ink);padding-top:13px}.section-bar h2{font-size:11px;letter-spacing:1.8px;text-transform:uppercase;margin:0;font-weight:700}.section-bar span{font-size:10px;letter-spacing:1px;color:var(--muted);text-transform:uppercase}.takeaways{display:grid;grid-template-columns:repeat(3,minmax(0,1fr));gap:25px;margin:0 0 38px;padding:0;list-style:none}.takeaways li{font-size:13px;line-height:1.65;border-right:1px solid var(--line);padding:0 23px 0 0}.takeaways li:last-child{border:0;padding-right:0}.takeaways strong{display:block;font-size:15px;line-height:1.35;margin-bottom:8px;letter-spacing:-.3px}.take-number{font-family:Georgia,serif;font-size:25px;color:var(--accent);display:block;margin-bottom:7px}.stories{display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:24px;margin:0 0 40px}.story{background:white;border:1px solid var(--line);padding:27px 28px 25px;border-radius:3px;min-width:0}.lead-story{grid-column:1 / -1;display:grid;grid-template-columns:minmax(0,1fr) 242px;gap:0 35px}.story-label{display:flex;align-items:center;gap:9px;text-transform:uppercase;letter-spacing:1.25px;font-size:9px;font-weight:700;color:var(--muted);margin-bottom:15px}.number{display:inline-flex;align-items:center;justify-content:center;width:25px;height:25px;border-radius:50%;background:var(--ink);color:white;font-size:10px;letter-spacing:0;flex-shrink:0}.announced{margin-left:auto;font-size:8px;letter-spacing:.6px;white-space:nowrap}.story h2{font-family:Georgia,'Times New Roman',serif;font-size:28px;line-height:1.15;font-weight:500;letter-spacing:-.8px;margin:0 0 18px}.lead-story .story-label{grid-column:1 / -1}.lead-story h2{font-size:37px;letter-spacing:-1.1px;grid-column:1;margin-bottom:19px}.report{font-size:14px;line-height:1.8;color:#42483e}.report p{margin:0 0 13px}.report p:last-child{margin-bottom:0}.report strong{color:var(--ink)}.report code{font-size:12px;overflow-wrap:anywhere;background:var(--soft);padding:2px 4px;border-radius:3px}.report a{font-size:12px;color:#636d50}.lead-story .report{grid-column:1;grid-row:3}.price-grid{grid-column:2;grid-row:2 / span 3;align-self:start;display:grid;gap:0;border-radius:3px;background:var(--soft);padding:14px 18px}.price-grid>div{padding:14px 4px;border-bottom:1px solid #d6dbc9}.price-grid span{display:block;font-size:9px;letter-spacing:1.3px;font-weight:700;color:var(--muted);margin-bottom:5px}.price-grid strong{font-family:Georgia,serif;font-weight:500;font-size:39px;letter-spacing:-1.5px;display:block;line-height:1.1}.price-grid small{font-size:10px;color:var(--muted)}.price-grid .highlight strong{color:#526333}.price-grid p{font-size:9px;line-height:1.6;color:var(--muted);margin:12px 4px 0}.our-read{border-left:2px solid var(--accent);padding-left:14px;margin-top:23px}.lead-story .our-read{grid-column:1}.read-label{display:block;font-size:9px;font-weight:700;letter-spacing:1.5px;color:#a9472d;margin-bottom:8px}.read-label i{font-family:Georgia,serif;text-transform:none;font-size:12px;letter-spacing:0;color:var(--muted);font-weight:400;margin-left:8px}.our-read p{font-size:13px;line-height:1.75;color:#5e6558;margin:0}.watch{background:var(--ink);color:#f6f7f2;padding:26px 29px 30px;border-radius:3px;margin-bottom:36px}.watch h2{font-size:11px;text-transform:uppercase;letter-spacing:1.8px;margin:0 0 22px}.watch ul{display:grid;grid-template-columns:repeat(3,minmax(0,1fr));list-style:none;gap:26px;margin:0;padding:0}.watch li{font-size:12px;line-height:1.75;color:#bfc6b6}.watch strong{display:block;font-size:14px;color:#f4f5ed;margin-bottom:8px}.watch strong::before{content:'↗';color:#ed8b69;margin-right:8px}.sources{margin-bottom:30px}.sources ol{padding:0;list-style:none;margin:0;display:grid;grid-template-columns:repeat(2,minmax(0,1fr));gap:18px 30px;counter-reset:source}.sources li{counter-increment:source;position:relative;padding-left:27px;font-size:11px;line-height:1.65;min-width:0}.sources li::before{content:'0' counter(source);position:absolute;left:0;top:0;font-family:monospace;color:var(--muted)}.sources a{display:block;font-weight:700;text-decoration:none}.source-url{display:block;font-size:9px;color:var(--muted);overflow-wrap:anywhere;font-family:Arial,Helvetica,sans-serif;line-height:1.6;margin-top:3px}.method{font-size:11px;line-height:1.8;color:var(--muted);margin:24px 0 0;padding-top:18px;border-top:1px solid var(--line)}.method strong{color:var(--ink)}footer{display:flex;justify-content:space-between;align-items:center;gap:18px;border-top:2px solid var(--ink);padding:21px 0 34px;font-size:10px;color:var(--muted)}footer .signoff{color:var(--ink);font-family:Georgia,serif;font-size:17px;font-style:italic}footer .brand{font-size:19px}
@media(max-width:720px){.page{padding:0 23px}.masthead{padding:22px 0 17px}.brand{font-size:20px}.hero{padding:28px 0}.hero h1{font-size:53px;letter-spacing:-2.3px}.hero-grid{grid-template-columns:1fr;gap:18px}.issue-note{border-left:0;border-top:1px solid var(--line);padding:12px 0 0;display:flex;gap:16px;flex-wrap:wrap}.issue-note strong{margin:0}.hero-lead{font-size:15px}.takeaways{gap:17px}.takeaways li{padding-right:14px;font-size:12px}.takeaways strong{font-size:13px}.stories{grid-template-columns:1fr;gap:18px}.lead-story{display:block}.story{padding:23px}.story h2,.lead-story h2{font-size:30px}.price-grid{grid-template-columns:repeat(3,minmax(0,1fr));gap:8px;padding:12px;margin:0 0 19px}.price-grid>div{border-bottom:0;padding:7px 4px}.price-grid span{font-size:8px;letter-spacing:.6px}.price-grid strong{font-size:30px}.price-grid p{grid-column:1 / -1;border-top:1px solid #d6dbc9;padding-top:8px;margin:0 4px;font-size:9px}.watch ul{grid-template-columns:1fr;gap:18px}.watch strong{margin-bottom:3px}.watch{padding:24px}.sources ol{grid-template-columns:1fr;gap:14px}footer{align-items:flex-start}.section-bar span{font-size:8px}.report{font-size:14px}}
@media(max-width:430px){.page{padding:0 17px}.hero h1{font-size:44px;letter-spacing:-1.8px}.masthead-right{font-size:8px;letter-spacing:.8px}.masthead-right span{font-size:8px;letter-spacing:0}.eyebrow{font-size:9px;letter-spacing:1.4px}.takeaways{grid-template-columns:1fr;gap:16px;margin-bottom:30px}.takeaways li{border:0;border-bottom:1px solid var(--line);padding:0 0 15px 37px;position:relative;font-size:13px}.takeaways li:last-child{padding:0 0 0 37px}.takeaways strong{font-size:14px;margin-bottom:4px}.take-number{position:absolute;left:0;top:0;font-size:24px}.story{padding:20px}.story h2,.lead-story h2{font-size:28px}.announced{font-size:7px}.story-label{font-size:8px;letter-spacing:.8px}.price-grid strong{font-size:27px}.price-grid small{font-size:8px}.price-grid span{font-size:7px}.hero-lead{font-size:14px}footer{flex-wrap:wrap}footer .signoff{font-size:15px}}
@media print{body{background:white}.page{max-width:none;padding:0}.stories{display:block}.story{break-inside:avoid;margin-bottom:20px}.lead-story{display:block}.price-grid{display:none}.report a{color:inherit}.watch{background:#f3f3ee;color:#23251f}.watch strong,.watch li{color:#23251f}}
</style></head><body><div class="page"><header class="masthead"><div class="brand">daily ai news<span class="star">✳</span></div><div class="masthead-right">Wednesday, September 30<span>2026 · A briefing from Stuff</span></div></header><main><section class="hero"><div class="eyebrow"><span class="dot"></span>The Wednesday edition · 5 stories</div><h1>From prompts<br>to <em>ongoing work.</em></h1><div class="hero-grid"><p class="hero-lead">Tuesday's announcements put ongoing agent work at the center of the AI business: a cheaper capable model, assistants that keep working between conversations, and shared documents where teams can collaborate with them. Meta is taking that pitch to small businesses, while Washington is asking the largest AI companies to strengthen their own oversight. Here are five developments announced September 29, checked against their sources this morning.</p><aside class="issue-note"><strong>The briefing window</strong><span>Announcements: Sept 29</span><span>Checked: Sept 30, 9:07 AM EDT<br>America/New_York · UTC−04:00</span></aside></div></section><section><div class="section-bar"><h2>Three things to know</h2><span>The short version</span></div><ol class="takeaways"><li><span class="take-number">01</span><strong>Watch the cost of completed work.</strong> Lower token prices help, but retries and human review still determine an agent's practical economics.</li><li><span class="take-number">02</span><strong>Connected work is the product.</strong> Dots, shared ChatGPT workspaces, and Muse's business integrations all compete to become part of an everyday workflow.</li><li><span class="take-number">03</span><strong>Separate commitments from enforcement.</strong> The new White House accord is a voluntary safety agreement; its implementation will matter more than its signing ceremony.</li></ol></section><div class="section-bar"><h2>The briefing</h2><span>Reported facts + our read</span></div><section class="stories"><article class="story lead-story"><div class="story-label"><span class="number">01</span><span>Models & economics</span><span class="announced">SEPT 29</span></div><h2>GPT-6.1 Sol arrives with cheaper cached context</h2><div class="price-grid"><div><span>INPUT</span><strong>$2</strong><small>per 1M tokens</small></div><div class="highlight"><span>CACHED INPUT</span><strong>$0.10</strong><small>per 1M tokens</small></div><div><span>OUTPUT</span><strong>$10</strong><small>per 1M tokens</small></div><p>GPT-6.1 Sol · Standard API pricing, per OpenAI</p></div><div class="report"><p><strong>Announced September 29.</strong> OpenAI released GPT-6.1 Sol in the API, ChatGPT Work, and Codex. Standard API pricing is <strong>$2 per million input tokens, $0.10 per million cached input tokens, and $10 per million output tokens</strong>. The API model identifier is <code>gpt-6.1-sol</code>. OpenAI says it approaches GPT-6 Astra on several coding, computer-use, and professional-work evaluations at one-fifth of Astra's standard input and output token prices. Those are vendor evaluations, not independently reproduced results. Availability includes eligible paid ChatGPT plans in Work and Codex; the announcement specifically says it is <strong>not yet available in Chat</strong>. <a href="https://openai.com/index/introducing-gpt-6-1-sol/">OpenAI's model announcement ↗ [1]</a></p>
</div><div class="our-read"><span class="read-label">WHY IT MATTERS <i>Our read</i></span><p>Agents that repeatedly consult the same documents or codebase can benefit from cheaper cached context. The useful comparison is quality and total cost on your own task set, including failed attempts, rather than a single benchmark score.</p>
</div></article><article class="story "><div class="story-label"><span class="number">02</span><span>Personal agents</span><span class="announced">SEPT 29</span></div><h2>Dots turns OpenAI's agent pitch into an ongoing relationship</h2><div class="report"><p><strong>Announced September 29.</strong> OpenAI introduced Dots, agents powered by GPT-6 Astra with their own cloud computers and access to connected apps. Rollout begins on Pro and Business Premium in eligible markets; enterprise users can try a beta when their administrator enables it. An important boundary: OpenAI says the proactive research that runs when users are not actively working with a dot uses <strong>read-only tools</strong>. That mode cannot send messages, change connected-app content, or control a browser or computer. Directed tasks have separate action and approval controls. <a href="https://openai.com/index/introducing-dots/">OpenAI's Dots announcement ↗ [2]</a></p>
</div><div class="our-read"><span class="read-label">WHY IT MATTERS <i>Our read</i></span><p>The product is moving from answering a prompt to retaining responsibility for work. That makes visible activity, clear access boundaries, and useful handoffs central to the experience. Always available should not be confused with unlimited permission to act.</p>
</div></article><article class="story "><div class="story-label"><span class="number">03</span><span>Shared knowledge</span><span class="announced">SEPT 29</span></div><h2>ChatGPT adds shared workspaces, documents, and richer plugins</h2><div class="report"><p><strong>Announced September 29.</strong> At DevDay, OpenAI introduced <strong>ChatGPT Space</strong> for teams and agents to build on shared knowledge, and <strong>Pages</strong> for collaborative documents. Space is available on Pro, Business, and Enterprise on desktop and web; mobile supports finding, reading, and sharing pages, with creation and editing still to come. Separately, plugin extensions let developers add sidebar destinations, interactive panels, and file viewers inside ChatGPT. <a href="https://openai.com/index/devday-2026-recap/">OpenAI's DevDay recap ↗ [3]</a></p>
</div><div class="our-read"><span class="read-label">WHY IT MATTERS <i>Our read</i></span><p>This expands the competition beyond chat interfaces into the places where teams keep their work. Developers gain another distribution surface, but should consider how customers export documents, retain permissions, and keep context usable across assistants. Integration depth and portability will increasingly need to coexist.</p>
</div></article><article class="story "><div class="story-label"><span class="number">04</span><span>Small business</span><span class="announced">SEPT 29</span></div><h2>Meta expands Muse into small-business operations</h2><div class="report"><p><strong>Announced September 29.</strong> Meta's Muse for Small Business adds skills and connectors to its existing agent, which is available in the US and Canada. The expansion connects Instagram professional-account analytics, Facebook Pages, and Meta ad accounts alongside services such as Shopify, QuickBooks, Canva, and Slack. Meta describes workflows spanning campaign planning, sales analysis, and financial reviews. It says publishing, sending, and spending require the user's approval. This is a concrete product expansion following the business-platform organization announcement covered yesterday. <a href="https://about.fb.com/news/2026/09/introducing-muse-small-business/">Meta's announcement ↗ [4]</a></p>
</div><div class="our-read"><span class="read-label">WHY IT MATTERS <i>Our read</i></span><p>Small businesses rarely need another empty chat window. They need help with the tools and records they already have. The test is whether connecting those systems produces dependable work with less supervision, especially when money or customer communications are involved.</p>
</div></article><article class="story "><div class="story-label"><span class="number">05</span><span>Policy & oversight</span><span class="announced">SEPT 29</span></div><h2>The White House secures a voluntary AI safety accord</h2><div class="report"><p><strong>Announced September 29.</strong> The White House and leaders from Google, Meta, Anthropic, SpaceXAI, NVIDIA, and OpenAI signed an accord addressing risks from advanced AI. The published text calls for internal controls and monitoring, internal review teams, independent external evaluations, and oversight by an independent board committee. It also calls for continued work on standards. These are commitments to an oversight process, not evidence that every safeguard is already operating. <a href="https://www.washingtonexaminer.com/news/white-house/4747747/full-trump-white-house-accord-ai-super-intelligence/">Accord text reproduced by the Washington Examiner ↗ [5]</a></p>
<p>AFP's reporting describes the agreement as voluntary and not legally enforceable. <a href="https://www.afp.com/en/trump-touts-ai-boss-pledge-self-regulate">AFP's report ↗ [6]</a></p>
</div><div class="our-read"><span class="read-label">WHY IT MATTERS <i>Our read</i></span><p>External evaluation could make safety claims more inspectable. The next questions are who performs the evaluations, what gets disclosed, and what happens when controls fail. A signed pledge alone does not answer them.</p>
</div></article></section><section class="watch"><h2>On our radar</h2><ul><li><strong>Real workloads:</strong> independent comparisons of Sol's reliability and total task cost.</li><li><strong>Rollout reality:</strong> which users can access the new agents and collaboration features, and where approvals interrupt or improve work.</li><li><strong>Accountability:</strong> concrete evaluation arrangements and disclosures following the safety accord.</li></ul></section><section class="sources"><div class="section-bar"><h2>Follow the sources</h2><span>6 source links</span></div><ol><li><a href="https://openai.com/index/introducing-gpt-6-1-sol/" target="_blank" rel="noopener noreferrer">OpenAI's model announcement ↗</a><span class="source-url">https://openai.com/index/introducing-gpt-6-1-sol/</span></li><li><a href="https://openai.com/index/introducing-dots/" target="_blank" rel="noopener noreferrer">OpenAI's Dots announcement ↗</a><span class="source-url">https://openai.com/index/introducing-dots/</span></li><li><a href="https://openai.com/index/devday-2026-recap/" target="_blank" rel="noopener noreferrer">OpenAI's DevDay recap ↗</a><span class="source-url">https://openai.com/index/devday-2026-recap/</span></li><li><a href="https://about.fb.com/news/2026/09/introducing-muse-small-business/" target="_blank" rel="noopener noreferrer">Meta's announcement ↗</a><span class="source-url">https://about.fb.com/news/2026/09/introducing-muse-small-business/</span></li><li><a href="https://www.washingtonexaminer.com/news/white-house/4747747/full-trump-white-house-accord-ai-super-intelligence/" target="_blank" rel="noopener noreferrer">Accord text reproduced by the Washington Examiner ↗</a><span class="source-url">https://www.washingtonexaminer.com/news/white-house/4747747/full-trump-white-house-accord-ai-super-intelligence/</span></li><li><a href="https://www.afp.com/en/trump-touts-ai-boss-pledge-self-regulate" target="_blank" rel="noopener noreferrer">AFP's report ↗</a><span class="source-url">https://www.afp.com/en/trump-touts-ai-boss-pledge-self-regulate</span></li></ol><div class="method"><strong>Sources & method.</strong> All five stories concern September 29 announcements; this edition contains no older carryover items. Product details come from opened company announcements. The policy summary uses the published accord text, reproduced by the Washington Examiner, with AFP reporting for corroboration. Company performance and safety statements are attributed, not independently validated here. Commentary labeled “our read” is interpretation. Prior editions were checked to avoid repeating unchanged headlines.</div></section></main><footer><span class="signoff">A small briefing for a very loud industry.</span><span>Made on <span class="brand">stuff<span class="star">✳</span></span></span></footer></div></body></html>