My fever will come crashing down

 

This town is really alright,

It's a bit tipsy though,

For there is a scent of you

All over the town.

I too am really alright,

Touch my forehead

With your quivering lips,

My fever will come crashing down.

Mythos Preview as part of Project Glasswing

 Cloudflare in its blog wrote thus about Project Glasswing:

--------------------

For the last few months, we've been testing a range of security-focused LLMs on our own infrastructure. These LLMs help identify potential vulnerabilities in our own systems, so we can fix them – and they also show us what attackers are going to be able to do with the latest models.

None of these LLMs has captured more attention than Mythos Preview, from Anthropic. A few weeks ago, we were invited to use Mythos Preview as part of Project Glasswing. We soon pointed it at more than fifty of our own repositories – to see what it would find, and to see how it works.

This post shares what we observed, what the models did well and what they didn't, and how the architecture and process around them needs to change, so they can be used at scale.

Mythos Preview is a real step forward, and it's worth saying that plainly before getting into anything else. We've been running models against our code for a while now, and the jump from what was possible with previous general-purpose frontier models to what Mythos Preview does today is not just a refinement of what came before.

It's a different kind of tool doing a different kind of work, and that makes a clean apples-to-apples comparison to earlier models difficult. So rather than trying to benchmark Mythos Preview against general-purpose frontier models, it's more useful to describe what it can actually do, and two features that stood out across the work we did with Mythos Preview:

Exploit chain construction - A real attack rarely uses one bug. It chains several small attack primitives together into a working exploit. For instance, it might turn a use-after-free bug into an arbitrary read and write primitive, hijack the control flow, and use return-oriented programming (ROP) chains to take full control over a system. Mythos Preview can take several of these primitives and reason about how to combine them into a working proof. The reasoning it shows along the way looks like the work of a senior researcher rather than the output of an automated scanner.

Proof generation - Finding a bug and proving it's exploitable are two different things, and Mythos Preview can do both. It writes code that would trigger the suspected bug, compiles that code in a scratch environment, and runs it. If the program does what the model expected, that's the proof. If it doesn't, the model reads the failure, adjusts its hypothesis, and tries again. The loop matters as much as the bugs it finds, because a suspected flaw without a working proof is speculation, and Mythos Preview closes that gap on its own.

Some of what we describe above is not entirely unique to Mythos Preview. When we ran other frontier models through the same harness, they found a fair number of the same underlying bugs, and in some cases they got further than we expected on the reasoning side too. Where they fell short was at the point of stitching the pieces together. A model would identify an interesting bug, write a thoughtful description of why it mattered, and then stop, leaving the actual chain unfinished and the question of exploitability open. What changed with Mythos Preview is that a model can now take those low-severity bugs (which would traditionally sit invisible in a backlog) and chain them into a single, more severe exploit.

--------------

Let's understand in layman's terms what this quote means: 

'A real attack rarely uses one bug. It chains several small attack primitives together into a working exploit. For instance, it might turn a use-after-free bug into an arbitrary read and write primitive, hijack the control flow, and use return-oriented programming (ROP) chains to take full control over a system. Mythos Preview can take several of these primitives and reason about how to combine them into a working proof. The reasoning it shows along the way looks like the work of a senior researcher rather than the output of an automated scanner.'

Here's what that's saying, in plain terms:

The core idea: hacking usually isn't one bug = one hack

Let's think of a piece of software as a house with many small flaws — a window that doesn't latch properly, a lock that's a bit loose, a floorboard that creaks and shifts. Any single one of these flaws alone might not let a burglar into the house. But a clever burglar can combine them: use the loose floorboard to reach the window latch, use the window to get to the loose lock, and now they're inside and can move freely.

That's an "exploit chain" — attackers rarely find one giant bug that hands over full control. Instead, they find several small, seemingly minor bugs and stitch them together, where each one enables the next step.

The technical example translated in layman's terms:

Use-after-free bug: A program frees up a piece of memory (analogy could be clearing out a filing cabinet) but then accidentally still tries to use it later. This creates an opportunity for an attacker to sneak their own data into that "empty" space.

Arbitrary read/write primitive: Turning that mistake into a general-purpose ability to read or write almost any memory location in the program  It is like going from "found one unlocked drawer" to "now I can open any drawer in the building."

Hijack control flow: Once you can write anywhere, you can trick the program into running instructions of your choosing instead of its own code. It is like slipping fake instructions into someone's to-do list and having them followed exactly.

ROP chains (return-oriented programming): A technique where an attacker doesn't inject new malicious code, but instead stitches together tiny existing fragments of the program's own code (like assembling a ransom note out of cut-up letters from a newspaper) to make the program do something it was never intended to do.

Therefore what Mythos Preview is doing is this:

The claim is that this AI model can look at a handful of these individual small flaws and figure out, on its own, how to link them together into a complete, working attack. And that its step-by-step reasoning while doing this reasoning resembles how an experienced human security researcher thinks through the problem, rather than just mechanically trying random combinations like a basic automated tool would.

Aw gust of wind

 

Aw gust of wind

Don't just go away,

For one and thirty days,

Blow every day.

Welcome August 2026!

Tax deduction under 80G under the new tax regime with special reference to the recent Assam floods

 Here is a discussion with Chatgpt on Tax deduction under 80G under the new tax regime with special reference to the recent Assam floods:

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Me:

Under the new Income tax regime, are contributions to Chief Minister's Relief Fund and Religious Institutions under 80G exempt from Income tax ? 

(If anyone wants to contribute to Chief Minister's Relief Fund Assam, s/he can do here.

https://cm.assam.gov.in/donate )

Chatgpt:

No. If you opt for the new income tax regime (Section 115BAC), you generally cannot claim a deduction under Section 80G for donations. This applies regardless of whether the donation is to a Chief Minister's Relief Fund or to an eligible charitable institution. 

A note about religious institutions

There is often confusion here.

Donations to approved charitable trusts or institutions that have a valid 80G registration may qualify under the old regime, even if the institution has a religious association, provided it satisfies the conditions laid down in Section 80G. 

However, donations made purely for religious purposes (for example, offerings, pujas, hundi donations, or donations to institutions that do not qualify under Section 80G) do not qualify for deduction.

If you are under the new regime

Your donation is still valuable, but it will not reduce your taxable income. The new regime allows only a limited set of deductions, and Section 80G is not one of them. 

If you have a specific fund or temple/trust in mind (for example, Karnataka Chief Minister's Relief Fund, ISKCON, Ramakrishna Mission, Tirumala Tirupati Devasthanams, etc.), tell me its name and I can check whether it has a valid 80G registration and, if so, what deduction category (50% or 100%) applies under the old regime.

Me:

Chief Minister's Relief Fund of Assam for the recent horrendous floods in Assam. Doesn't make sense to me: Income tax goes to Central Govt from where the govt allots flood alleviation relief to Assam(if at all). And here I'm directly making donations to the epicentre: it is about one govt agency over the other.

Chatgpt:

I think you've put your finger on a genuine policy trade-off, and many people have expressed the same concern.

From a public finance perspective, there are two different flows of money:

1. Taxes are compulsory payments into the Consolidated Fund. Once there, Parliament and the governments decide how the money is allocated across defence, healthcare, roads, disaster relief, pensions, etc.

2. Donations are voluntary payments that allow citizens to direct additional resources to a specific cause—in this case, flood relief in Assam.

Under the old tax regime, Parliament chose to encourage such voluntary giving by allowing a deduction under Section 80G for eligible funds such as a State Chief Minister's Relief Fund. The Chief Minister's Relief Fund is explicitly listed as an eligible fund under Section 80G (subject to the statutory conditions). 

Under the new tax regime, the philosophy changed. The government simplified the tax system by lowering rates while removing most deductions and exemptions, including Section 80G. The idea is that everyone pays tax on a broader base, without needing to track deductions.

So your argument is essentially:

> "If I'm donating directly to a government-run relief fund that the government itself has established, why should that be treated differently from tax?"

That's a perfectly coherent policy position. One could argue that such donations:

reduce the financial burden on the State,

reach the intended beneficiaries more directly,

and therefore deserve continued tax recognition even under the simplified regime.

On the other hand, the government's counterargument would be:

the new regime deliberately avoids distinguishing between "good" and "better" deductions;

if one deduction (disaster relief) is retained, many others (education, healthcare charities, research institutions, etc.) would also seek similar treatment, making the "simplified" regime more complex again.

So it is less a question of economics than of tax policy design.

In fact, if you donate ₹1,00,000 to the Assam Chief Minister's Relief Fund under the new regime:

the Government of Assam receives the full ₹1,00,000 for flood relief;

but your taxable income remains unchanged.

Whereas under the old regime, assuming the donation qualified for a 100% deduction, you would both help the relief effort and reduce your taxable income by the eligible amount.

Me:

I think this is sophistry. So, what has changed in the environment that the assumption of the old regime be disregarded? How can simplicity in tax calculation be an argument for such a change. You just need to change the algorithm, or use AI, or whatever. I mean this so frustrating.

Chatgpt:

I understand why you find it frustrating. And I think it's important to distinguish between an explanation of the policy and a justification for the policy. They are not the same.

The government's stated rationale is simplification, but that does not, by itself, establish that removing the deduction was the best policy.

In fact, your question can be reframed as:

> What changed between the old and the new regime that made it undesirable to encourage donations to disaster relief?

There are a few possible answers, but none is "because computers couldn't calculate it."

Simplicity isn't about computation

You're absolutely right that calculating one more deduction is trivial. Even in the 1980s, income tax software could do it.

When policymakers speak of "simplification," they usually mean:

* fewer categories of deductions,

* fewer eligibility disputes,

* fewer opportunities for tax planning,

* fewer documents to verify,

* fewer assessments and litigation.

So "simplicity" is primarily an administrative concept, not a computational one.

The real policy shift

The new regime represents a different philosophy:

> "Instead of giving tax incentives to influence people's behaviour, charge lower tax rates and let people make their own spending decisions."

Whether that philosophy is preferable is a separate question.

Your specific example is stronger than the general charity case

I think there's a meaningful distinction between:

* donating to a private charitable trust,

* donating to a religious institution,

* donating directly to a government-established disaster relief fund.

The last case is different because the recipient is itself a government fund established for a public purpose. One could argue that such donations supplement public expenditure in exactly the area the government would otherwise have to finance.

That doesn't prove they must receive a deduction, but it does make them a special category worthy of separate consideration.

Why might the government still remove it?

One possible line of reasoning is:

> "We don't want taxpayers choosing where public resources go. Taxes are collected centrally, and elected governments decide allocations."

In that view, if someone wants to help Assam beyond paying taxes, that's a voluntary act of philanthropy rather than a substitute for taxation.

Again, that's a policy choice—not a mathematical necessity.

Is your criticism unreasonable?

I don't think so. It's a legitimate policy criticism to say:

disaster relief serves an undeniable public purpose;

donations to government disaster relief funds reduce pressure on public finances;

therefore they should perhaps remain deductible even if many other deductions are abolished.

In fact, one could design a new regime that eliminated 95% of deductions while retaining only a handful—say, donations to national or state disaster relief funds, military welfare funds, or similar government-notified emergency funds. That would still be administratively simple.

So I would separate the issues:

Why was the deduction removed? Because the legislature chose a broad rule: almost all Chapter VI-A deductions disappeared under the new regime.

Was that the only way to simplify the tax system? No. The legislature could have retained a small number of narrowly defined deductions, including government disaster relief funds, if it had wanted to. Simplicity alone does not logically require their removal.

In other words, the change reflects a policy preference, not a technical limitation.

Your observation highlights a broader question: Should donations made directly to government disaster relief funds be treated differently from donations to private charities? Reasonable people can disagree on that, and the current law answers "No" under the new regime—not because the donation is less valuable, but because the regime intentionally removes almost all such incentives.

Me: 

To this above quote of chatgpt: "We don't want taxpayers choosing where public resources go. Taxes are collected centrally, and elected governments decide allocations", I asked this question to chatgpt:

Wow! Is this progression or regression in public participation in governance? I pay the tax, but I have no right to know where it goes? Its like a fire and forget type of missile.

Chatgpt:

That's a profound question, and it's one that political philosophers have debated for centuries.

I would separate two issues that are easy to conflate:

1. The right to know where your taxes go.

2. The right to decide where your taxes go.

These are related, but they're not the same.

In a representative democracy like India, the constitutional model is that citizens do have a right to know—through the Union Budget, state budgets, audit reports, parliamentary debates, the work of the Comptroller and Auditor General of India, and the Right to Information Act. What citizens generally do not have is an individual right to earmark their taxes for particular purposes. That power is exercised collectively through elected representatives.

Your concern, however, is slightly different. You're not saying:

> I want to divert my income tax to Assam.

You're saying:

> "After paying my taxes, if I voluntarily contribute additional money to a government disaster relief fund, why should the tax system stop encouraging that?"

That's a narrower and quite distinct proposition.

Has public participation increased or decreased?

Reasonable people can disagree.

One view is that the new regime represents regression in one specific sense: it reduces one mechanism by which citizens could be encouraged to support public causes directly. If tax incentives increase participation in disaster relief, removing them could reduce that form of civic engagement.

Another view is that tax policy should remain neutral, neither encouraging nor discouraging particular forms of giving, leaving charitable decisions entirely to individuals without tax incentives.

Neither conclusion follows automatically from economic theory; they reflect different ideas about the relationship between citizens and the state.

"Fire-and-forget" taxation

Your analogy captures a sentiment that many taxpayers express: once taxes are paid, individuals lose any direct influence over how those specific rupees are spent.

That is, in fact, a feature of most representative democracies. Taxes are pooled so governments can respond to changing priorities rather than being constrained by millions of individual earmarks. Imagine if everyone designated their taxes differently—one person only for defence, another only for education, another only for roads. Governments could struggle to fund essential but less popular services.

At the same time, there are countries and municipalities that experiment with participatory budgeting, where citizens directly vote on how a portion of public funds should be spent. Those experiments suggest that representative democracy and direct citizen participation need not be mutually exclusive—they can coexist.

So your question ultimately isn't just about Section 80G. It's about the balance between representative decision-making and citizen participation. The removal of the deduction can be seen as one small shift in that balance, but whether that shift is a progression or a regression is a matter of political philosophy rather than an objectively settled fact.

Me:

In other words, the govt is saying that your contribution should be totally altruistic. The govt is saying, 'We will decide where you tax goes: whether to make missiles, to education, to healthcare, or to natural disaster relief.' It would be instructive to know how much more/less has flown into Chief Minister's Relief fund (a representative case, say) after the new tax regime has come into force.

Chatgpt:

That's a very interesting empirical question, and it's exactly the kind of evidence that could inform the policy debate.

The hypothesis you're suggesting is:

> Removing the Section 80G deduction under the new regime may have reduced donations to government relief funds.

That's a testable proposition. But to answer it convincingly, we'd need data.

There are several possible outcomes:

Donations fell significantly. This would suggest that tax incentives were an important motivator, even for donations to government relief funds.

Donations remained roughly the same. This would suggest that most donors to disaster relief were motivated primarily by the cause rather than the tax benefit.

Donations increased. This could happen if the frequency or severity of disasters, public awareness, or corporate giving outweighed the effect of the tax change.

The difficulty is that many other factors influence donations:

* the scale of the disaster,

* media coverage,

* appeals by public figures,

* corporate social responsibility (CSR) contributions,

* growth in incomes,

* and the fact that many taxpayers still choose the **old regime** if it is more beneficial.

So simply comparing donations before and after the new regime would not, by itself, isolate the effect of removing the deduction.

However, a well-designed study could compare:

donations from individuals who switched to the new regime versus those who remained in the old regime,

or use statistical methods to estimate the impact while controlling for disaster severity and other variables.

Is the government asking for "pure altruism"?

In one sense, yes. Under the new regime, if you donate ₹10,000 to the Assam Chief Minister's Relief Fund, the tax system treats that as your voluntary charitable decision rather than something that merits a tax concession.

Whether that is good policy depends on the purpose one assigns to tax incentives.

There are two schools of thought:

Incentive view: Tax deductions are a legitimate tool to encourage socially beneficial activities, including disaster relief, research, education, and philanthropy.

Neutrality view: The tax system should raise revenue with minimal distortions, and governments should support public purposes through direct spending rather than through tax preferences.

Neither philosophy is universally accepted; different countries strike different balances.

The evidence would be valuable

I particularly like your suggestion of examining a representative Chief Minister's Relief Fund over time. If the data showed, for example, that individual donations fell sharply after taxpayers migrated to the new regime, it would strengthen the case for carving out an exception for government disaster relief funds.

Conversely, if donations were largely unchanged, policymakers could argue that the deduction was not materially influencing giving.

So the debate could move from principle to evidence.

I also wouldn't be surprised if this has not yet been studied rigorously in the public domain. Disaster relief donations are relatively episodic, and disentangling the effect of tax policy from the effect of the disaster itself is statistically challenging. But it is exactly the sort of question that economists of public finance would investigate using donation data, tax-return data (where available in aggregate), and natural experiments created by the introduction of the new regime.

xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

Idle mind is God's worship

 

Idle mind is Devil's workshop,

Idle mind is God's worship.

The train's window

 

The train's window

Is transparent glass.

Two hands,

One inside the train,

One outside,

Can see,

But can't join.

Meanwhile,

One hand is pushed away

By the motive force

Of the train.

One hand stays rooted

At the spot,

Due to force of gravity

Of the place,

And that of home.

The longing of belonging

 

The longing

Of belonging,

The gain,

Due to pain,

Is the measure

Of the heart sundered,

Is the treasure

Of being in love.