BUSINESS

Indiana Fever star point guard Caitlin Clark, a three-time WNBA All-Star selection, is playing in this year’s game after missing out in 2025 due to a groin injury.

Are autonomous vehicles like Waymo really safer than human drivers?That’s the question some of the biggest tech companies in the world are investing billions of dollars to answer. Alphabet owns Waymo, Amazon owns Zoox, and Tesla has its own Robotaxi division.The advantages robots have over human drivers are apparent. Robotaxis don’t drink and drive, don’t drive while tired, don’t drive distracted, and don’t get vengeful when someone cuts them off.Half of U.S. states reported discernible blood alcohol concentration for at least 70% of fatally injured passenger-vehicle drivers in 2023, according to the Insurance Institute for Highway Safety (IIHS). So it’s clear that eliminating drunken driving would significantly improve road safety. And clearing the roads of people whose driving ability is impaired for other reasons puts us well on our way to a transportation utopia.On the other hand, driverless vehicles have displayed some disturbing patterns as they accumulate more miles on the road, leading to some awkward interactions with human drivers and even some dangerous situations with children and pedestrians.So the IIHS conducted a study to find out once and for all: Is Waymo safer than human drivers?Waymo cars were involved in nearly 70% fewer crashes than humansWaymo driverless vehicles deployed in San Francisco, Phoenix, Los Angeles, and Austin were involved in 68% fewer crashes than human drivers per vehicle mile traveled (VMT), according to a new study by the Insurance Institute for Highway Safety.“The results show that, on a limited scale, these driverless cars are safer than human drivers — who can be impaired or drowsy or suffer lapses in attention,” IIHS President David Harkey said.Researchers estimated that only 22% of the 736 public-road crashes in which automation was engaged between 2021 and 2024 were likely police-reportable, including 89 crashes involving Waymo vehicles. Waymo’s crash involvement was 76% lower in Phoenix, 35% lower in San Francisco, and 71% lower in Los Angeles, but was 4% higher in Austin, though the sample size there was relatively small.Waymo vehicles were involved in 85% fewer single-vehicle crashes and 81% fewer injury crashes per VMT.But before futurists declare the discussion about safety over, the study came with a major caveat: “However, the present data collection system isn’t good enough to allow continuous monitoring of a large-scale expansion.”The data used in the study go back to 2021, when the National Highway Traffic Safety Administration began requiring self-driving cars on public roads to report crash involvements that result in fatalities, injuries, or property damage of any severity.But while companies are required to report crash data, they are not obligated to report the miles they log or whether such miles involved a human driver. Waymo provides that information voluntarily, but its competitors do not, making a holistic assessment impossible. “Even with the mileage data that Waymo provides, a direct comparison with the crash rates of humans is impossible without time-consuming operations to account for the differences between the reports required by the SGO and what humans report,” the IIHS said. Meanwhile, human drivers are not required to report crashes that result in no injuries or less than $1,000 in property damage to the police. According to IIHS, about half of all crashes and a third of injury crashes by humans go unreported. Waymo says that the IIHS report validates the company’s own analysis of the first 56.7 million miles of fully autonomous driving under its belt. Waymo’s own number showed its vehicles were involved in 79% fewer injury crashes compared to human drivers. “We welcome this new research from IIHS, which confirms our previous peer-reviewed analyses and reinforces the significant safety benefits of the Waymo Driver,” a Waymo spokesperson told TheStreet. “We also support the authors’ recommendation for more robust reporting requirements for AV operators — requirements that Waymo already voluntarily meets.”

A new IIHS study noted that the present data collection system isn’t good enough to draw firm conclusions on the safety of autonomous vehicles versus human drivers.Josep LAGO / AFP via Getty Images

U.S. senators question Tesla FSD safety dataIn May, Reuters reported that Tesla was exaggerating its safety claims for FSD and that it is using a team of “data labelers” to help improve the AI that powers FSD.Based on this revelation, Elon Musk’s declaration that FSD is already up to 10 times safer than human drivers and ready for more widespread adoption may ring hollow.So Senators Edward Markey (D-Mass.) and Richard Blumenthal (D-Conn.) sent a letter to the National Highway Traffic Administration saying the Reuters report exposes “dangerous gaps” in its autonomous vehicle data collection.“Tesla has repeatedly told investors, consumers, and the public that FSD is far safer than human driving, but the data analysis justifying those claims is weak and misleading. These representations are not merely marketing claims; they may shape how drivers use Tesla’s FSD, how the public understands the risks of the technology, and how regulators evaluate potential safety defects,” the letter stated.According to the letter, Tesla’s data justifying the “10 times safer” claim is flawed for several reasons, including:Comparing unlike crash outcomes that made Tesla look betterComparing newer Tesla vehicles to the entire U.S. vehicle fleetCounting FSD-involved crashes only if it is active at the time of crash or within five seconds. The NHTSA uses a 30-second time threshold for all ADAS systemsRelying on incomplete automated telemetryTesla is cooking the books, according to the senators, and the NHTSA has not been able to get the real data it needs, which makes the whole situation more dangerous for drivers.“The push to allow more autonomous vehicles on public roads depends heavily on the claim that these driving systems are safer than human drivers,” the letter stated.“To the extent that Tesla or other vehicle manufacturers are misleading the public about their safety data, however, consumers may choose to purchase or ride in an AV based on the unproven expectation that they are safer than non-autonomous vehicles. This type of information asymmetry is a classic market failure, which will likely result in more AVs on the road — and potentially more traffic injuries and fatalities if those systems are not in fact as safe as they claimed.”Currently, the NHTSA does not require vehicle manufacturers to submit data on the number of vehicles they operate, the distances they travel, and other data that could help contextualize crash rates.They say that is the type of data that “would help prove or disprove Tesla’s safety claims.”For this reason, the senators are asking the agency to “significantly expand autonomous vehicle data reporting requirements.”Related: Waymo shows it learned critical lesson from previous blackouts

Whatever you might expect from a graphic novel by actor John Cusack, ‘Momo’ is certain to surprise you with its range of offbeat influences.

The U.S. Department of Defense has entered an aggressive new era of spending this year. Before launching into its massive new tech overhauls, it kicked off the momentum with a $266 million firm-fixed-price contract awarded to Rocket Lab for 12 suborbital launches. But aerospace isn’t the only sector cashing in on this modernization push; legacy enterprise software is also securing massive wins.Oracle spent the first half of 2026 fighting a narrative about debt. The stock lost roughly a third of its value between January and July as investors questioned whether the company could afford its AI data center buildout, according to The Motley Fool.On Thursday, July 23, Oracle landed a 10-year, nearly $7 billion contract with the U.S. Department of Defense, and the market barely blinked.That gap between the size of the deal and the size of the reaction is the real story. It says less about Oracle’s government business, which is healthy, and instead highlights what investors actually worry about heading into the back half of the year.Consolidating a decade of Defense spendingThe Pentagon awarded Oracle a 10-year Indefinite Delivery/Indefinite Quantity contract on July 23, according to a press release distributed through the agency’s Enterprise Software Initiative.The base value covers $3.31 billion over the first five years, rising to $6.99 billion if the department exercises all five option years, the release states.The agreement covers on-premises Oracle software across every military branch, the Coast Guard, and the intelligence community, CNBC reported.Related: Analyst sends chilling Oracle stock verdictIt was negotiated by the Department of the Navy as a non-competitive direct award and marks the department’s first ever direct contract with Oracle for its on-premises tools.Oracle’s relationship with the department is not new. The CIA was Oracle’s first customer decades ago, CNBC noted, and the company has sold into the military since the 1990s. What changed is the structure, not the relationship.Scale explains why that structure matters. The Department of Defense employs more than 3.4 million civilians and service members across dozens of agencies and branches.Before this contract, that workforce bought Oracle software through scattered, one-off purchases instead of a single framework.

Oracle secured a 10-year Pentagon software contract worth up to $6.99 billion, but shares rose only 2% to 3% as debt concerns continued to weigh on the stock.Douglas Rissing / Getty Images

Wall Street’s muted reaction reveals what investors are pricingOracle (ORCL) shares rose about 2% to 3% in extended trading on Thursday, July 23. For a contract that can reach $7 billion, that is a modest move.Context explains why. Oracle just closed its worst week since the 2001 dot-com bust in late June, falling 19% in five trading days as investors focused on its balance sheet, CNBC reported.The company was carrying about $130 billion in debt as of late May, with capital spending up 162 percent to nearly $56 billion for the fiscal year, CNBC found.Oracle has also said it plans to raise another $40 billion through debt and equity in fiscal 2027 to keep funding data centers, Motley Fool reported.A government contract, however large, does not resolve that math. It adds a predictable revenue stream to a story that has otherwise been about unpredictable, debt-funded AI capacity, and investors appear to be treating the two as separate questions rather than one offsetting the other.A direct access to government advantageThe Pentagon signed a similarly sized enterprise software agreement in May, but the comparison is not to Oracle’s biggest AI rivals.That deal, worth $9.69 billion over five years, covers Microsoft 365 and Azure licensing, and it was awarded to Dell Federal Systems, not Microsoft, CNBC reported at the time. Dell manages the licensing relationship on Microsoft’s behalf.More Oracle:Mizuho sends intriguing Oracle stock message after 38% dropOracle stock makes rattling move after major setbackOracle stock falls for a seventh session as filing risk landsOracle’s contract skips that layer entirely. The department is buying straight from Oracle, a structural difference that gives Oracle more direct control over how the relationship evolves, more visibility into how its software actually gets used, and more of the margin that would otherwise go to a reseller.That distinction matters beyond this one contract. It suggests Oracle’s decades of entrenched, unglamorous enterprise relationships still carry weight the newer AI infrastructure story does not fully capture.Department of Defense’s fewer, bigger software betsThe Department of Defense has been consolidating its technology purchasing all year, from the Dell-Microsoft agreement in May to a January initiative aimed at accelerating commercial AI adoption across the military.The pattern favors vendors already embedded in government systems over newer entrants trying to break in.For Oracle, that is a quieter kind of advantage than a splashy AI partnership, but a more durable one. Software running inside classified systems is not swapped out easily, regardless of what happens to a company’s stock price.This contract only covers on-premises software, the part of Oracle’s business furthest from the AI data center buildout worrying investors.The next test is whether Oracle can win a similarly structured deal for its cloud and AI products, not just its legacy databases, as defense agencies decide which vendors they trust to run the next decade of military infrastructure. That answer will say more about Oracle’s future than the July 23 contract does.Related: Oracle stock makes rattling move after major setback

It’s official! Interview with the Vampire season 4 is a go at AMC. It’s titled Queen of the Damned and the reins of Akasha’s story have been handed to Hannah Moscovitch.

Two tankers loaded with Saudi crude turned around in the Red Sea on July 22. They had been heading toward China and India. The Houthis had just claimed strikes on them. Around the same time, Kazakhstan’s oil exports were falling again after the Caspian Pipeline Consortium suspended loadings at its Black Sea terminal because of attacks. Two separate disruptions, two separate shipping corridors, on the same day.Goldman Sachs was already working on a new oil research note. Lead analyst Daan Struyven and his team had been watching the same headlines. By the time the note published, Goldman had a clear message for investors: The price forecast stays at $80 Brent for Q4 2026, according to Investing.com — but the risk around that forecast has shifted. The chances of going higher have increased. The chances of going lower have not.What Goldman Sachs said about oil prices and Red Sea shipping risksThe two events on July 22 are exactly the kind of thing Goldman’s note was built around. Houthi forces claimed strikes on two Saudi oil tankers transiting the Red Sea. The tankers turned around before reaching the Bab-al-Mandab Strait, according to CNBC. At the same time, CPC oil loadings appear to have declined following fresh attacks on tankers at its Black Sea terminal, affecting the pipeline that carries about 80% of Kazakhstan’s crude exports to global markets.Neither development on its own changes the supply picture fundamentally. Together, Goldman says, they’re the kind of incremental pressure that tilts near-term risk to the upside without yet being large enough to move the baseline.Related: U.S. blocks Strait of Hormuz: Here’s what’s next for oil pricesOil flows through the Bab-al-Mandab Strait, the chokepoint connecting the Red Sea to the Gulf of Aden, have averaged nearly 9 million barrels per day over the past 30 days. That includes roughly 4 million barrels per day that would be extremely hard to reroute if disruptions hit the Bab-al-Mandab, the Strait of Hormuz, and the Suez Canal simultaneously. The Houthis declared a maritime embargo against Saudi Arabia, threatening to cut off the kingdom’s Red Sea oil exports entirely, according to CNBC. Saudi loadings at the Yanbu port have stayed stable at around 5 million barrels per day for now. Whether they stay that way depends on whether the Houthis follow through.Why Goldman kept its $80 Brent oil forecast unchangedGoldman didn’t raise its forecast because the baseline still assumes geopolitical tensions gradually ease before Q4. That assumption is doing a lot of work in the model. The bank expects Middle Eastern production to fall in the second half of 2026 as a result of the conflict, which provides support. But two things are pulling the other way.Middle Eastern output came in stronger than expected in June, adding supply the market hadn’t fully priced in. And Goldman downgraded its demand expectations for China, South Korea, and the Middle East after weak consumption data from all three. More Oil & Gas:Drivers face an unpleasant surprise at the gas pumpU.S. blocks Strait of Hormuz: Here’s what’s next for oil pricesA big shift in the U.S. energy market is about to happenThat demand softness takes some pressure off prices even as supply risk rises on the other side.WTI stays at $76 for Q4 2026 under the same unchanged baseline. For 2027, assuming Hormuz stays open, Goldman forecasts Brent at $75 and WTI at $70, with a global surplus of 3.2 million barrels per day. That’s not a bullish picture for next year. But Goldman doesn’t think it translates to a price collapse either, as TheStreet reported.Goldman’s oil price upside scenario and how bad it could getGoldman outlined what happens if the baseline assumption is wrong. If the Strait of Hormuz remains significantly disrupted through 2027, Brent could exceed $120 in Q4 2026 and average around $100 throughout 2027. If disruptions spread simultaneously to the Bab-al-Mandab and the Suez Canal, Goldman estimates an additional $25 of upside on top of that scenario. Gulf oil production would only fully recover by December 2027 in that case, with pipeline extensions helping bridge the gap.There’s also a downside. Brent could fall to the low $60s by the end of 2027 if supply exceeds Goldman’s expectations and demand losses prove more persistent. That scenario looks less likely right now, but it’s there.Goldman also gave investors a specific trade to go with the view. If you want to hedge against persistent Mideast and Russia disruptions, the bank said to go long the December 2026 to March 2027 European diesel/gasoil timespread. That’s the instrument Goldman thinks captures the geopolitical risk most efficiently right now.

Governments are expected to add roughly 1.2 million barrels per day of demand through strategic stockpiling in 2027.Ronaldo/Getty Images

Why Goldman expects oil prices to stay firm through summer 2026Even under the base case, Goldman expects oil to hold most of its recent gains through July and August. Global visible oil stocks have already hit a year-to-date low. Three things are keeping them under pressure.Middle Eastern production is expected to fall in July. Summer travel should add roughly 0.8 million barrels per day of demand in Q3 versus Q2. And strategic petroleum reserve releases have slowed sharply after heavy drawdowns in Q2, with South Korea and Japan now actively rebuilding reserves rather than releasing them. That shift removes a source of supply that had been quietly cushioning the market.Put those together and you get a summer where inventory keeps drawing even if the broader economic backdrop stays mixed.What prevents oil from collapsing even in a 2027 surplusGoldman’s 2027 surplus forecast is 3.2 million barrels per day, assuming Hormuz stays open. That’s a large number. Normally it would drag prices down hard. Goldman doesn’t think it will, and the reasons are specific.Governments are expected to add roughly 1.2 million barrels per day of demand through strategic stockpiling in 2027. U.S. shale economics create a natural floor: sustained prices below roughly $60 a barrel slow drilling and reduce future supply, which tightens the market from the supply side. Both factors together mean Brent probably doesn’t fall much below the high-$60s even in the surplus scenario, as TheStreet reported.Goldman’s bottom line is that the balance of risks points up, not down, especially near-term. The official forecast hasn’t moved yet. But the conditions that would move it higher are building.Related: JPMorgan resets oil price target for rest of 2026

Every industry builds a room where it keeps the dangerous thing. Chemical plants have containment vessels. Banks have vaults.Artificial intelligence (AI) labs have sandboxes, sealed computing environments where a model can be pushed to its limits without touching anything real.The rule is simple. Whatever happens inside the sandbox stays inside the sandbox.Testing a model’s hacking ability makes that rule load-bearing, because the test only works if you switch off the safety refusals that would normally stop the model cold.So the lab builds the tightest box it can, strips the guardrails, points the model at a target, and measures what it does next.The results usually surface months later as a benchmark score in a research paper, discussed at conferences by people who speak in acronyms.Nobody outside the labs pays much attention, and for about three years, the arrangement has held well enough that nobody needed to.It stopped holding this month.OpenAI disclosed on July 21 that a combination of its own models chewed through containment during an internal evaluation, reached the open internet, and broke into Hugging Face, the unaffiliated platform where much of the world’s open-source AI is hosted.Neither company is publicly traded. That has not stopped the disclosure from landing on the desk of every chief information security officer with a budget.Why AI security spending is already a board-level problemStart with the money, because the money explains the reaction.Worldwide end-user spending on information security reached $213 billion in 2025 and is projected to rise 12.5% to about $240 billion in 2026, according to Gartner.More Artificial Intelligence:Workers over 55 in Al-exposed jobs face new realityVisa hands banks an edge against their rivals with Al toolNvidia CEO doubles down on Al and stock market verdictThat is healthy growth for the sector. It is also a rounding error next to what the same companies are spending to buy and deploy AI in the first place.Federal officials have been circling that gap for months. The pattern was already visible: regulators treating machine-speed attacks as a financial stability question rather than an IT question.Most enterprise security stacks were built to catch a human intruder, or a script written by one. They assume an attacker gets tired, makes noise, and works a shift.What changed on July 21 is that the alternative stopped being a forecast.

OpenAI’s models escaped a sandbox and hacked Hugging Face during a cyber benchmark test.Europa Press News / Getty Images

What OpenAI disclosed about the Hugging Face breachThe sequence matters more than the summary.Hugging Face went public first. The company said on July 16 that it had detected and contained an intrusion into part of its production infrastructure, one driven end-to-end by an autonomous agent system.At that point, nobody knew whose agent it was.Related: OpenAI just admitted something that has the AI industry on edgeFive days later, OpenAI identified the attacker as itself. The models involved were GPT-5.6 Sol and a more capable pre-release model, both running with cyber refusals reduced for evaluation purposes, according to OpenAI.Here is the part that keeps me up. The models were not trying to cause damage. They were trying to pass a test.Told to solve a cyber-capability benchmark called ExploitGym, they spent enormous compute finding a way out instead. They located a zero-day flaw in a software package proxy, escalated privileges across the research network, reached a machine with internet access, then reasoned that Hugging Face probably hosted the benchmark’s answers.They were right, and they went and took them.The company described the event as an “unprecedented cyber incident, involving state-of-the-art cyber capabilities,” according to OpenAI.Cheating on a test is a very human motive. Doing it by finding a previously unknown software flaw at three in the morning is not.The timeline behind the AI breach numbersThe published record is thin but specific.On July 16, Hugging Face disclosed unauthorized access to internal datasets and service credentials.More than 17,000 attacker events were reconstructed by the company’s own analysis agents, according to Hugging Face.July 21, OpenAI attributed the intrusion to its own models under evaluation.A previously unknown vulnerability in a package proxy provided the path to the open internet, OpenAI also indicated.Information security spending is forecast at roughly $240 billion for 2026, according to Gartner.Those five lines describe a failure mode for which no current security vendor sells a finished product.The guardrail problem nobody priced into cyber stocksThen came the detail I did not expect, and it is the one investors should sit with.Hugging Face said that when it tried to analyze the attack using commercial frontier models, the requests “were blocked by the providers’ safety guardrails.” Feeding real exploit payloads to a hosted model looks identical to attacking with one.So the defenders ran their forensics on an open-weight Chinese model, GLM 5.2, hosted on their own hardware.Read that again. The attacker was bound by no usage policy. The defenders were.That asymmetry is a product roadmap for every security vendor on the market, and the sell side has noticed. Palo Alto Networks (PANW) and CrowdStrike (CRWD) have both been repriced this year around agentic AI defense, and Microsoft (MSFT), OpenAI’s largest corporate backer, sells the security tooling that sits underneath much of the enterprise cloud.Lawmakers noticed, too. Rep. Greg Casar (D-Texas) called the disclosure alarming, saying “AI is developing extremely fast with no real regulations to keep us safe,” according to Al Jazeera.Congress has spent two years arguing about AI copyright and AI trade secrets. This is the first incident that hands it a security question with a named victim.What the Hugging Face breach means for your portfolioIf you own an S&P 500index fund, you own this problem twice.You own the companies building models that can now chain novel exploits without ever seeing the source code. You also own the companies selling the defense, whose addressable market just expanded by a category that did not exist in last year’s budget.For the next several quarters, I would watch three things rather than the headlines. Whether security vendors report accelerating deals tied specifically to agentic threats. Whether frontier labs publish containment standards that an outside auditor can actually check. Whether Washington converts alarm into a disclosure requirement with teeth.There is a smaller, more personal item, too. Hugging Face advised users to rotate access tokens and review recent account activity, which is the same hygiene that protects your brokerage login and your email.The uncomfortable takeaway is not that a model went rogue. It did not. It followed instructions with a literalism nobody had priced in, and the shortest path to a passing grade ran straight through another company’s production database.That behavior will not stay inside test environments. The next system that does it will not have a lab publishing a blog post about it afterward.Related: Tech expert predicts an OpenAI collapse

“I would likely be the one responsible for dealing with the house.”

The change in tariff policies that went into effect Friday is distinctly worrying for some small firms, but other say it’s business as usual.

Tyson Fury stopped Mariusz Wach in seven rounds in Thailand in his final tune-up, setting up a long-awaited heavyweight showdown with Anthony Joshua this fall.