In case you hadn't worked it out - the sun's image is a disc with a diameter of approximately 1/110th of the lens focal length. With a wide angle lens, this is quite small, and any heating produced locally can more easily be conducted away, preventing serious damage.

Conduction is constant regardless of the size of the sun's image, though; it is also isotropic. Most of the heat gets conducted out the back of the chip and into the package regardless of the image size. A big image may make a few percent or so difference, but I have reason to doubt that this would result in crossing some damage threshold.

Moreover, any fairly minor change of direction that the lens is pointed in will mean a different area is capturing the sun's image. With a telephoto, the area heated increases and may not be so easily conducted away.

This makes sense for shutterless cameras. But few if any of them can zoom to such a long focal length that the sun takes up a big percentage of the sensor area. And even if they did, most of these cameras are f/5 to f/9 at the long end.

Meanwhile, on the shuttered camera side, nobody is going to be shooting the sun with the camera set to bulb.

Finally, any photographer smart enough to be concerned about sensor damage by the sun is also going to be using a tripod, right? ;-)

I really doubt the real-life applicability of the shaky-camera principle; as Steven said, I wonder if anyone has first hand experience with this.

Also, the reflectivity of the surface on which the sun is imaged affects how much of the energy is absorbed - hence you can burn a hole in a black coated shutter although it isn't quite at prime focus.

I think another important physical quantity at issue here is the emissivity. For CCD and CMOS sensors this quantity is around 0.88 at the d-Line (wavelength 587.56nm). So it should dump heat rapidly through radiation.

The other thing to consider is reactivity and how that scales with temperature. For fabric shutters, reactivity (primarily oxidation) is very high and does not require a large temperature increase to accomplish. For titanium shutters, reactivity is rather lower. But in CCD and CMOS sensors, it is almost non-existent. The materials are relatively inert. The basic mechanism of heat-mitigated damage to CCD and CMOS devices is through recrystallization. Looking at materials tables, I gather that temperatures around 1,000 degrees C, sustained for some minutes or possibly hours, is required to recrystallize these sensors. This is the reason for the skepticism I express above.

Real-life experience backs this up. One of the SOHO mission's camera shutters was jammed open for several hours while staring at the sun. Despite being exposed not only to light, but also to radiation far more energetic than visible light (such as extreme ultraviolet), this produced no damage except where an active region's image happened to be falling on the sensor. Even then it did not destroy the sensor or render it in any way inoperable.

So I am skeptical that anyone has damaged their sensor taking a picture of the sun. Some alternative explanations for instances of apparent sensor damage may be that anti-reflective coatings on the near-IR exclusion filter were burned (or perhaps a plastic IR exclusion filter was melted); that the glue that is found in some CCD and CMOS packages was melted; that other electronics near the sensor were destroyed through conductive overheating; or even that something volatile inside the chamber got heated and subsequently condensed onto near-sensor surfaces. (This latter is the typical form of solar damage to spacecraft star trackers, I am told.) These are fairly mundane repair issues, though, not garbage collector issues like real sensor damage.