Russia "Cannot Achieve Land War Success" As Adaptability Favors Ukraine

The increased frequency and brutality of Russia's terror attacks on civilian Ukrainian targets are because Putin understands that he can no longer win the land war and may even fail to preserve some, perhaps much, of what his troops initially took. 

The Ukrainians have now launched two counteroffensives this past spring and summer - around Lyman, Dnipropetrovsk and Zaporizhzhia which have retaken hundreds of kilometers of territory. They have done so with an innovative combined arms maneuver warfare approach that has, to some degree, neutralized Russian drone forces while degrading logistics and reinforcement as well as enabling the reintroduction of infantry in armored vehicles. Russia simply has no answer for this. JL

Phillips O'Brien reports in his substack:

Zelensky believes the Russian air attacks are increasing because they “cannot achieve success” in the land war. The Ukrainians have been liberating territory near Lyman - up to 250 square miles of territory some say - but are keeping it quiet for operational security (the Ukrainians have become very good at that—they only will announce successes once a campaign phase is over), as they did earlier this summer when they took back a chunk of land at the center of the line. This shows the Ukrainians can advance without taking mass casualties even as Russian losses remain shockingly high. The Ukrainians operate with small numbers of well-trained, motivated forces, in close coordination with masses of drones. It is about being as precise as possible, knowing where a soldier must go and what he might encounter on the way. It is the opposite of the Russian strategy of wasting mass manpower. The adaptability balance in the land war continues to favor Ukraine. 

10,000 Ukrainian AI Hornet Drones Deployed Over 3 Months To Attack Russia

This is how autonomy, at further range and lower cost but with greater range - and all at scale - is impacting the war. 

Ukraine has deployed 10,000 AI-enabled medium range (200 km) strike drones for $5,000 a piece over three months. They are a big part of the reason why Ukraine's combined arms assaults in Lyman, Dnipropetrovsk and Zaporizhzhia have succeeded as the Russians are unable to defend a much greater area behind the front, allowing Ukraine to degrade their operational capabilities. JL

Shawn Reed reports in Medium:

Ukraine has deployed over 10,000 Hornet strike drones over three months to attack targets within Russia. It reflects Ukraine’s rate of scaling up the use of medium-range, AI-assisted drones. Not only does it attain a range of over 200 kilometres, but it has onboard systems that can help track their target when maintaining a continuous link to an operator may become difficult. It is particularly useful against logistics, fuel supplies, military infrastructure, and other targets in the rear areas. Thousands of drones equipped with a range of over 200 kilometres, onboard navigation, targeting assistance and a low, $5,000 unit cost alters the situation as Russia is forced to defend a much larger area. The Hornet is important because it is more autonomous, with further range, cheaper, and far more numerous.

"Meatgrinder With Kremlin Providing the Meat" As Russian Pressure Shrinks

Reports across all sectors of the front reveal a uniform shrinking of Russian pressure on Ukrainian positions. This suggests both that the Kremlin is running low on resources - especially the manpower whose lives they so profligately disregard - and that the Ukrainian counteroffensives are so disrupting Russian plans that they are being forced on the defensive.

That the war has become 'a meatgrinder with the Kremlin providing the meat,' as a Ukrainian commander offered, is an indication that, on the frontline, the Russians no longer have the initiative, nor are their earlier conquests necessarily safe. JL

Stefan Korshak reports in the Kyiv Post:

The trend of fighting seems to be shrinking Russian pressure in almost all sectors. Based on the field descriptions this is the textbook “we’ve just about run out of men to send on attack” phase. This is the fifth (!) Russian summer offensive when we’ve seen this happen. In the Zaporizhzhia sector, attacks are still taking place but significantly decreasing and with less-numerous groups, by “about 40 percent.” In Pokrovsk - Dobropillia, the ratio of Ukrainians to Russians lost can be 1 to 60. The Russians are killed or wounded, do not gain ground. At Konstyantynivka, Russian attacks have fallen off to very little. And at Lyman, it’s official, 3rd Corps completed its counter-offensive liberating 10 villages and recapturing150 square kilometers of Russian territory. The war right now is running a meatgrinder with the Russians providing the meat. 

Is the Current AI Safety Meltdown Also About the AI Profitability Crisis?

The Wall Street Journal last week called the rising concerns about AI's existential threat 'the great American freak-out,' with an evident expression of disdain for those 'who don't get it' putting concerns about extinction above the profits to be made by a shrinking cadre of tech oligarchs and their inside investors. 

But others are raising a different specter: that, in fact, the leaders of Anthropic, OpenAI et al are jumping on the 'slow it down' bandwagon because they now see that there is no way of operating and earning their way out of the financial apocalypse their heedless pursuit of being first has caused. None of the AI firms are close to profitability, nor do they see a real path for getting there. And all the major tech firms will be cash-flow negative next year, which is exactly as economically troubling as it sounds. So slowing things down gives them a plausible excuse for doing what common sense should have already dictated, which is that the AI arms race is built on increasingly shaky projections so jumping off that bandwagon may actually save their companies. 'May' being the operative word. JL

Shira Ovide reports in the Washington Post and Amrith Ankumar and colleagues report in the Wall Street Journal:

All over the country, this was the week that panic about the existential threat of AI leaped from internet debates and Silicon Valley to spread across mainstream America. It is a backlash that has been building for months. Questions about when, if ever, will this AI payoff arrive are growing more urgent. AI costs and doubts are spreading. The U.S. stock market has swooned this summer over fear of the AI bubble going bust. The boom is pushing up inflation, adding to challenges in tackling affordability. Financial results show AI’s mammoth costs are largely swamping the sales boost: next year, Google, Amazon, Microsoft, Meta and Oracle are projected to have negative free cash flow. At Google, for every dollar of cash generated in the past three months, $1.15 paid for AI chips and data centers. "It's not clear these guys are going to make money on it," a bet so gargantuan is cannot possibly pay off.

As Russia's Invasion Falters, Putin Boasts of Imaginary Victories

In a pattern with strategic implications, Vladimir Putin is increasingly making claims of battlefield success which are immediately disproved by evidence from independent and Ukrainian sources. 

The desperation inherent in these false achievements comes as Putin tries to maintain that Russia is still a powerful military to reckoned with, despite the considerable evidence to the contrary. In addition to the Russian army, air force and navy's battlefield failures, his dependence on China and North Korea for support to continue the war undermines those pretensions to imperial greatness. In effect, Russia has become a client state whose performance is only useful as a model for the designs of its more powerful allies. And since Putin well knows that his sometime pal President Trump 'does not like losers' he has to do everything he can to prevent that label from sticking to him. JL

Peter Dickinson reports in The Atlantic Council:

On September 1, Putin proudly announced that his troops had captured Sviatohirsk. Less than a week later, Ukrainian commander Andriy Biletsky posted a video address in the center of Sviatohirsk demonstrating that the city was still very much under Ukrainian control. “To Russia’s military and political leadership. Take off your clown noses and stop making fools of yourselves.” Putin’s habit of exaggerating Russia’s battlefield gains reflects his mounting unease over the credibility gap between the Kremlin’s superpower pretensions and his army’s underwhelming performance in Ukraine. With the invasion now well into its fifth year, Russia currently controls less than 20% of Ukraine and is still fighting over villages within walking distance of the original front lines in February 2022. For a country that entered the war with a reputation as the world’s second most powerful military, this a remarkably poor return.

In First Ever Battle Of Naval Drones, Ukraine's Destroys Russia's

In what may be the most significant naval engagement of the modern era, a Ukrainian unmanned naval drone destroyed a Russian naval drone in the Black Sea yesterday.

Military historians may cite the introduction of planes taking off from ships or of submarines hunting surface ships at scale, but this is more reminiscent of the battle between the Union ironclad Monitor against the Confederate ironclad Merrimac in 1862, during the American Civil War, which is considered the beginning of the modern era of naval warfare as wooden hulled sailing ships began to be replaced by armored steam or coal-powered ships. This may be the start of the unmanned naval ship era. JL

Tamara Nowakowsky reports in the Kyiv Independent:

A Ukrainian unmanned surface vessel destroyed a Russian naval drone in the Black Sea in what Ukraine's Navy described as the "first-ever naval battle" between unmanned boats. The operation was carried out on Sept. 12. The Russian drone was destroyed by a Ukrainian Sargan-3000 unmanned boat, equipped with a 12.7-mm Protector remote weapons station, according to the Navy. The drone-on-drone encounter marked a historic first. The joint operation represents a shift from naval drones attacking conventional ships to drone-on-drone maritime combat. Both Ukraine and Russia  are increasingly moving toward unmanned warfare at sea.

Growing List of AI Hacks Could Be Human Culture Failure Infesting AI Firms

Suddenly, AI executives are falling all over themselves insisting that AI 'be slowed down' for safety reasons. That all of these firms also happen to remain massively unprofitable with almost no legitimate means of achieving financial stability is a problem for another post, though it suggests that they are going to come calling on Washington for a taxpayer-funded bailout. But with the number and scale of AI agents 'escaping' and attacking other platforms while a growing number of  extremely well compensated AI employees are quitting to sound the alarm about AI's danger to humanity, the biggest crisis maybe the culture inside AI companies and the pressure they are getting from investors.

This issue is not a technical one. It is a human factor problem and, specifically, the result of selfish, irresponsible and greedy leadership placing power and money above safety. The AI escapes were known internally, but employees were either afraid to say anything or were ignored - or threatened - when they did. Because the scale and money were more important than the potential consequences. That may be changing, although as mentioned above, probably more for financial reasons than out of any newly acquired sense of concern for the public. JL

Grace Huckins reports in MIT Technology Review:

Last month's AI security incidents, in which OpenAI (and later, Anthropic) agents escaped and hacked into other AI platforms while trying to cheat on a test is a wild story. (But) employees noticed what was happening at multiple points—and either failed to raise the alarm or were not heard when they did. “When you look at (such) incidents, people try to find the technical source of failure, but that can give a very misleading sense of why the failure occurred. If people are cutting corners all the time, if they are not in a culture that prioritizes safety and has appropriate incentives and structures, [accidents] are bound to happen.” Big alignment problems exist in the disconnect between company culture and the public interest. And as tough as technical AI research might be, fixing those problems could prove far harder.