> daily_signal(2026_07_24)

The FAA cleared over 1,000 police agencies to fly drones out of sight, on a form no city council sees. Abbott turned on his own data centers, ChatGPT Business cannot export, DOE named 278 projects.

PickBits Daily Signal · Friday, July 24, 2026

1. More than 1,000 US police agencies now hold FAA waivers to fly drones beyond the operator's line of sight — the agency issued more of them between April 2025 and February 2026 than in the previous seven years combined, according to an EFF analysis published July 23.

The Drone-as-First-Responder model is simple to describe and hard to undo: a police department mounts drones on fixed rooftops around a city, and when a 911 call comes in, a drone launches automatically and arrives overhead before a patrol car does. The bottleneck was never the hardware — it was the FAA rule requiring an operator to keep the aircraft in visual line of sight, which meant one human standing outside per drone. In April 2025 the FAA streamlined the Part 91 waiver process for Beyond Visual Line of Sight operations, and the dam broke. EFF's July 23 analysis puts the count at more than 1,000 agencies holding waivers as of February 2026, against 976 waivers issued in total across the seven years from 2018 through April 2025 — meaning the agency issued more in ten months than in the preceding seven years. The operational consequence is the part worth sitting with: BVLOS is what turns a drone program from a staffing cost into a force multiplier, because one operator in a room can supervise several automated aircraft over a city. The civil-liberties problem is not that a drone responds to a 911 call; it is that the aerial-surveillance capability is acquired through a federal waiver process with no public-comment step, no local vote, and no requirement that a city council know it happened. Departments that could never have won budget approval for a helicopter now field a persistent aerial platform at a fraction of the cost, and the record of what those drones capture on the way to a call — over back yards, protests, and everyone who is not the subject of the call — is governed by whatever retention policy the department wrote for itself, if it wrote one. Read this as a capability-expansion story rather than a discrete-harm story, because that is what it is: nothing new happened to a specific person this week. What happened is that the acquisition curve went vertical while the oversight curve stayed flat.

Key fact: CHECK YOUR OWN TOWN BEFORE ASSUMING THIS IS SOMEONE ELSE'S PROBLEM — EFF's Atlas of Surveillance is a free searchable database of which US agencies have acquired which surveillance technologies, organized by city and county: https://atlasofsurveillance.org/. With more than 1,000 agencies holding BVLOS waivers as of February 2026, 'my department probably doesn't have one' is now a bad assumption. If your agency is listed, the three documents to request are the BVLOS waiver itself (a federal record naming the agency, its approved operating area and conditions), the Drone-as-First-Responder standard operating procedure, and the 12-month flight log. The gap between the waiver's approval date and the first public mention of the program is, in most of these cities, the whole story.

EFF Deeplinks analysis published 2026-07-23: more than 1,000 law-enforcement agencies held FAA Beyond Visual Line of Sight (BVLOS) waivers as of February 2026 — the approvals that make Drone-as-First-Responder programs operationally viable · The rate change, quoted from EFF: the FAA 'issued more waivers between April 2025 and February 2026 than it had in the previous seven years combined' — against 976 total waivers granted from 2018 through April 2025 · The regulatory trigger: the FAA streamlined its Part 91 waiver process in April 2025, removing the requirement that an operator keep the aircraft in visual line of sight and allowing a single operator to supervise multiple automated drones launched from fixed locations across a city · The oversight gap EFF identifies: the BVLOS waiver process has no public-comment requirement and no local-approval step, so a department can acquire persistent aerial-surveillance capability before a city council votes on — or is told about — the acquisition · primary source

2. Continuing 05-13 #3: the Texas data-center fight jumped from one county to the whole state — Gov. Greg Abbott, who announced Google’s $40 billion Texas investment, has now ordered regulators to keep data centers from raising Texans’ electricity bills and promised to strip their state tax breaks, while Ohio voters gather signatures for a statewide referendum to ban new data-center construction outright.

The AP's Marc Levy reported on July 23 what local officials have been watching for a year: opposition to AI data centers has stopped being a zoning fight and become an electoral force, and it is splitting the rural Republican base that usually anchors red-state majorities. The clearest tell is Greg Abbott. Abbott championed the buildout and personally announced Google's $40 billion Texas investment; Texas is projected by JLL to pass northern Virginia as the world's largest data-center market by 2030, and Google is building four data centers in northwest Texas including a facility near Amarillo billed as the world's largest. Abbott has now reversed: he ordered regulators to ensure Texans do not pay higher electricity bills because of data centers, promised a legislative agenda to regulate them, said he would eliminate their state tax breaks, and proposed blocking construction in 'rural Texas neighborhoods.' His Democratic opponent, Gina Hinojosa, calls that last phrase 'doublespeak that makes no sense' and has made the issue central: 'They are owned by the richest men in the world. We're all footing the bill. There are no rules. It is the wild west of data centers.' The pattern repeats across states. In Ohio, Sherrod Brown is running television ads calling Sen. John Husted 'the face of data centers in Ohio' while citizens gather signatures for a statewide referendum to ban data-center construction. In Wisconsin, Democratic gubernatorial candidate Francesca Hong pledged to 'stop AI data centers.' In Pennsylvania, Gov. Josh Shapiro is taking fire from his own side, with Democrat Sam Burleigh accusing him of 'allowing these data centers to come in and take over acres and acres of prime farmland.' What makes this different from a normal partisan issue is that the opposition is not partisan. Texas Agriculture Commissioner Sid Miller, a Republican, opposes the projects. The Lubbock-area protest organizers, Stephen Sanders and Hallie Bertrand, are Republicans who say they may not vote for Abbott but are not sure they can vote for Hinojosa either. Giles Dalby, a Republican county commissioner and cattle rancher, worries data centers will get priority in water rationing in a region where shortages already hurt the sugar-cane industry. Clayton Tucker, the Democratic nominee for state agriculture commissioner, calls it 'the most cross-partisan issue I've ever seen.' For anyone planning AI capacity, the operative fact is that the binding constraint on the buildout is shifting from power and chips to local political consent — and consent is now being withdrawn at the state level, by both parties, in the states where the capacity was going to go.

Key fact: IF YOUR ELECTRIC BILL WENT UP AND YOU LIVE NEAR A PROPOSED DATA CENTER, THE COST-ALLOCATION DOCKET IS WHERE THAT GETS DECIDED, NOT THE ZONING HEARING — the question that determines whether a data center raises residential rates is whether your state's utility regulator assigns the cost of new transmission and generation to the large-load customer or spreads it across the ratepayer base. Abbott's order to Texas regulators is exactly this instrument. Find your state's public utility commission docket for large-load or data-center tariffs and file a comment; these proceedings are open, sparsely attended, and decide the bill. In Ohio, the parallel lever is the signature drive for the statewide construction-ban referendum.

Associated Press, Marc Levy, 2026-07-23: opposition to AI data centers is now shaping statewide campaigns in Texas, Ohio, Wisconsin, Pennsylvania, Arizona and New York, and is dividing the rural Republican voters who normally counterweight Democratic urban strongholds · Texas Gov. Greg Abbott's reversal: after championing the buildout and announcing Google's $40 billion Texas investment, Abbott ordered regulators to ensure Texans do not pay higher electricity bills because of data centers, promised a legislative agenda to regulate them, said he would eliminate their state tax breaks, and proposed blocking construction in 'rural Texas neighborhoods' · Democratic Texas gubernatorial nominee Gina Hinojosa, quoted: 'They are owned by the richest men in the world. We're all footing the bill. There are no rules. It is the wild west of data centers.' She says she would block data-center permits unless projects benefit Texans and residents have input, and calls Abbott's 'rural Texas neighborhoods' framing 'doublespeak that makes no sense' · Ohio: Democratic nominee Sherrod Brown is airing television ads attacking Republican Sen. John Husted as 'the face of data centers in Ohio,' while citizens gather signatures for a statewide referendum to ban data-center construction outright · Wisconsin Democratic gubernatorial candidate Francesca Hong pledged to 'stop AI data centers'; in Pennsylvania, Democrat Sam Burleigh accuses Democratic Gov. Josh Shapiro of 'allowing these data centers to come in and take over acres and acres of prime farmland' · The opposition is cross-partisan: Texas Agriculture Commissioner Sid Miller (R) opposes the projects; Lubbock-area protest organizers Stephen Sanders and Hallie Bertrand are Republicans unsure they will back either candidate; Republican county commissioner and cattle rancher Giles Dalby warns data centers could get priority in water rationing where shortages already hit the sugar-cane industry; Democratic agriculture-commissioner nominee Clayton Tucker calls it 'the most cross-partisan issue I've ever seen' · Scale context (JLL, cited by AP): Texas is projected to overtake northern Virginia as the world's largest data-center market by 2030; Google is building four data centers in northwest Texas, including a facility near Amarillo described as the world's largest · primary source

3. The ChatGPT tiers companies pay the most for are the ones that cannot export their own chats: OpenAI gives Free, Plus, Pro and some Edu users a chat export, but not Business or Enterprise workspaces — whose admins get only 30 days of Compliance Platform logs — so a freelance journalist published a scraper on GitHub this week to get his data out.

The Register reported on July 23 that OpenAI's chat-export feature is available to Free, Plus, Pro and some Edu customers, but not to ChatGPT Business or Enterprise workspaces. That is the inversion worth sitting with: the individual paying $20 a month can take a full archive of their conversations with them; the company paying per seat for an entire workforce cannot. What Enterprise admins get instead is access to OpenAI's Compliance Platform, and that is retention-limited to 30 days — an audit and eDiscovery surface, not a portability mechanism. OpenAI offered no justification for the split and did not respond to questions about the rationale. Into that gap stepped Conrad Quilty-Harper, a freelance journalist, who published a tool called Scrapemychats on GitHub this week. It drives a logged-in browser session to walk the conversation list and archive each thread with its attachments into offline-browsable HTML, running slowly because OpenAI throttles the requests. Quilty-Harper is candid that this is a workaround rather than a right: 'It's probably a breach of their terms of service. But hey, their entire business model is based around scraping copyrighted material.' For anyone who owns an organization's software inventory, the substance here is not the tool — it is that a system of record grew inside the company without the exit properties a system of record needs. Two or three years of an engineering org's design discussions, an support team's customer context, a legal team's drafting history now live in a workspace whose only sanctioned extraction path expires after 30 days. That is a vendor lock-in fact and a records-retention fact at the same time, and it is the kind of thing that gets discovered during a migration or a litigation hold rather than during procurement. It also sits oddly against data-portability expectations: consumer tiers get portability that paid enterprise tiers do not, which is the reverse of how every other enterprise SaaS category has settled.

Key fact: IF YOU OWN THE SOFTWARE INVENTORY, ANSWER ONE QUESTION THIS WEEK: CAN WE GET OUR CHATGPT WORKSPACE DATA OUT, AND IN WHAT FORMAT? For Business and Enterprise the answer today is that there is no bulk export and the Compliance Platform keeps 30 days. Write that down as a known gap in your application inventory alongside the tools that do have export, because the moment it matters — a migration, an acquisition, a vendor change, a litigation hold — is the moment it is too late to negotiate. Verify against your own tenant rather than assuming your plan differs: https://www.theregister.com/ai-and-ml/2026/07/23/openai-wont-let-some-customers-export-their-chats-but-this-tool-will/5277795

The Register, 2026-07-23: OpenAI's chat export is available to Free, Plus, Pro and some Edu customers, but ChatGPT Business and Enterprise workspace users have no export access — the paid organizational tiers, not the consumer ones, are the ones locked out · What Enterprise admins get instead: access through OpenAI's Compliance Platform, which is limited to 30-day retention — an audit and eDiscovery surface rather than a data-portability mechanism · OpenAI provided no justification for the restriction and did not respond to The Register's inquiries about the policy's rationale · The workaround: freelance journalist Conrad Quilty-Harper published a tool called Scrapemychats on GitHub this week that drives a logged-in browser session to retrieve and archive conversations with their attachments into offline-browsable HTML; downloads run slowly because OpenAI throttles the requests · Quilty-Harper on the tool's legal standing, quoted: 'It's probably a breach of their terms of service. But hey, their entire business model is based around scraping copyrighted material.' · primary source

4. Energy Secretary Chris Wright named the first 278 Genesis Mission projects on July 22 — $293 million across 342 institutions, 168 of them university-led — and NIH stood up a Bio Genesis Mission the same week, opening its pediatric-cancer data ecosystem to DOE supercomputers to train models across hundreds of rare cancer subtypes.

The Genesis Mission is the federal government's attempt to point AI at science rather than at chat, and this week it stopped being an executive order and started being awards. On July 22, Energy Secretary Chris Wright announced the first 278 projects selected under DOE's Genesis Mission funding call — $293 million, drawing 342 participating institutions: 16 DOE and NNSA national laboratories, 142 universities, 157 companies and 13 nonprofits. By award leadership the split is 87 national-lab-led, 168 university-led, 19 company-led and 4 nonprofit-led, which is worth noting because it inverts the concentration pattern most federal AI money has followed — the majority of these awards went to universities, not to labs or contractors. DOE called it the largest response to a funding opportunity in its history. The named science areas are nuclear energy (including the largest single award, a three-year $60 million nuclear-energy investment), critical-mineral extraction, intelligent chip design and commercial fusion. Awardees get access to the Genesis Mission Platform — AI agent frameworks, industry-supplied models and software, and DOE high-performance computing — which is the part that matters most for a university group that could never otherwise buy that compute. The health half runs through NIH, which launched its Bio Genesis Mission with more than $1.2 billion in obligated FY2026 and planned FY2027 funding, against six challenges: modeling complex biological systems, domestic biomanufacturing, biosecurity and biosurveillance, accelerating pediatric cancer research, shortening drug discovery and clinical translation, and chronic-disease root causes. The concrete mechanism there is a data move rather than a model announcement: HHS is providing its integrated pediatric-cancer data ecosystem and its national network of cancer centers so DOE supercomputers can train and apply models across hundreds of rare cancer subtypes — the exact problem where each individual subtype has too few patients to support a conventional study and pooling is the only route to a signal. NIH's stated goal is to double the pace of biomedical innovation from discovery to health impact within five to ten years. Two honest caveats belong with the good news: 'selected for award negotiations' is not 'funded and running,' and the White House OSTP science blueprint released the same week drew a mixed reaction from researchers for weighting AI over life-sciences funding.

Key fact: IF YOU ARE A UNIVERSITY RESEARCHER OR AT A NATIONAL LAB, THE COMPUTE ACCESS IS THE ASSET TO CHASE, NOT JUST THE GRANT — Genesis Mission awardees receive access to the Genesis Mission Platform: AI agent frameworks, industry-supplied AI models and software, and DOE high-performance computing. With 168 of the 278 awards university-led and DOE reporting the largest response to a funding opportunity in its history, this is now the main federal on-ramp to frontier-scale compute for academic science. Watch for the next RFA cycle and the platform's access terms rather than waiting for a standalone compute program: https://www.energy.gov/articles/secretary-energy-chris-wright-announces-first-genesis-mission-projects-selected-accelerate

US Department of Energy, 2026-07-22: Secretary of Energy Chris Wright announced the first 278 projects selected under the Genesis Mission Request for Applications, which DOE described as 'the largest response to a funding opportunity in DOE history' (energy.gov/articles/secretary-energy-chris-wright-announces-first-genesis-mission-projects-selected-accelerate). The $293 million total is stated in DOE's companion release the same day, 'Energy Department Announces $293 Million in Funding to Support Genesis Mission National Science and Technology Challenges' (energy.gov/articles/energy-department-announces-293-million-funding-support-genesis-mission-national-science) — the announcement page itself names only the largest single award, a three-year $60 million nuclear-energy investment · Award leadership breakdown: 87 led by DOE and NNSA national laboratories, 168 led by universities, 19 led by companies, 4 led by nonprofit organizations — across 342 participating institutions (16 national labs, 142 universities, 157 companies, 13 nonprofits, 14 other) · Named science areas: nuclear energy, critical-mineral extraction, intelligent chip design and commercial fusion energy; the largest single project is a three-year $60 million nuclear-energy investment. Awardees receive access to the Genesis Mission Platform — AI agent frameworks, industry-supplied AI models and software, and DOE high-performance computing resources · Secretary Wright, quoted: 'The 278 projects selected today represent the very best of our nation's scientific enterprise' · NIH Bio Genesis Mission (nih.gov/bio-genesismission), announced July 2026 alongside more than $5 billion in total federal Genesis Mission commitments unveiled 2026-07-22: more than $1.2 billion in obligated FY2026 and planned FY2027 NIH funding against six challenges — modeling complex biological systems with AI, domestic biomanufacturing, biosecurity and biosurveillance, accelerating pediatric cancer research, shortening drug discovery and clinical translation, and chronic-disease root causes · The pediatric-cancer mechanism: HHS is providing its integrated pediatric cancer data ecosystem and its national network of cancer centers so that DOE supercomputers can train and apply models across hundreds of rare cancer subtypes; NIH's stated goal is to double the pace of biomedical innovation from discovery to health impact within five to ten years · primary source